Correct me if I am wrong but doesn't Java's sandbox model refuses to load untrusted program into memory (if set up o only run signed applets) and restricta a user to a subset of available functions for manipulating rights-managed data?
As far as I know there isn't anything in Java that distinguishes the access rights of any particular piece of data, but you can install a custom SecurityManager in the JVM that can deny certain actions taken by particular threads, use a custom ClassLoader to ensure that signed classes can take extra privileges not granted by default, and ensure only signed classes get access to rights-managed data. Unless it's in silicon, you can still break the JVM, a la Ken Thompson's famous login/cc hack.
115.2k isn't the data rate. At 115kbaud, the stop bit (or is that start bit?) is just ignored noise and you send either 0x00 or 0xff if you want to send a high or low signal.
The basic rate for the 1-Wire bus is 14.4kbps. The overdrive rate for the 1-Wire bus is 115.2kbps. Regardless of which speed is used, assuming a 16550 with FIFO high water mark set to 13 bytes, the CPU is still potentially servicing over 1000 very general-purpose interrupts per second, on a none-too-fast machine. If it weren't for the pesky ACK bit one could turn the UART bit rate down to 14.4kbps, easily enough done with an 8250 descendent.
These days one may as well use a USB-to-1-Wire bridge. The chip presumably exists or will soon (DS2490, IIRC).
Where I work sells a small micro that hooks up to 4 channels and mounts in a nice rackmount box and sends the data out a 9600 baud serial line. I've got a small program that read teh data, shoves it in/tmp so mrtg can pull it out every 5 minutes. I also know when people leave the computer room door open.
That's not so bad. I almost wrote code for a PIC to do some basic 1-Wire stuff but couldn't find great code to handle the bus and had to roll up my electronics workbench rather hastily due to a sudden lack of rent money. A friend of mine had his whole house 1-Wired for temperature based on a slightly more powerful machine in the 486 family. I wonder if he's gotten around to implementing control yet -- he had a handful of DS2406 switches but was pretty much incompetent with a soldering iron.
Did that part of the article really ring true for anybody else?
Yes, yes it did, being an outcast of sorts myself. It fits well with the common tale of the insanity of the "sane", for one thing, and having attempted to learn real-life dating I can't see for the life of me how anyone sane would put themselves through that ridiculous Masonic handshake and basket of expectations just to be partnered for the night. It was all so much easier when it was non-verbal.
ALICE is nothing more than a bunch of preprogrammed responses to common statements and questions, what the
hell is the big deal about that?
The big deal about that is that preprogrammed responses to common statements and questions are a huge part of human conversation, or less generously that human conversation is mostly useless filler.
The more I read/. the more I find Wallace's misanthropy rubbing off on me.
and is multidrop by design. It's pointless to put each probe on its own bus and design interface circuitry yadda yadda, unless you have so many probes that 750ms to poll each one is too much of a wait.
That said, a 33MHz machine might not have the UARTs and the speed to handle 115.2kbps data, which is the 1-Wire high-speed rate.
I set up a similar system using Dallas' TINI Tiny InterNet Interface to poll the temperatures in three parts of my studio apartment: the fridge[1], the hallway to the bathroom, and the kitchen. On the TINI I ran a server in Java that, under control from any TCP client, would continuously poll the 1-Wire bus and work TTL I/O that was on the TINI sockets board but unpopulated[2]. The client for this was a perl program that polled the TINI, put data into a database[3], and switched a relay on/off which controlled a loud muffin fan[4] based upon the temperature in the main room. I had a servlet on my app server which would draw graphs using gdJava[5] from the database and a JSP which would allow zooming in and out.
Yes, I agree it was grossly overcomplicated, and if I were doing the same thing today I'd probably have had the TINI post data to the app server instead, so as to cut the perl script out of the loop. But I didn't really trust the TINI as much more than a really smart 1-Wire interface, so in reality I'd probably design with the RS232-based 1-Wire interface card instead, or use a PIC to do all this.:-)
-jhp
[1] It was a bar fridge and the freezer frequently affected the thermostat due to their closeness. In line with programmer virtue #1, I wanted to wait to defrost until the temperature in the fridge put my food at risk. [2] This was my first experience with surface-mount soldering. The new rev of the sockets board has a lot less cool stuff on it now -- the LCD interface is gone, for one. [3]The TINI's Java environment wasn't hefty enough to handle PostgreSQL JDBC drivers so something else had to do this. [4]It was 2001. I was trying to be nice to the electric grid and my own power bill, even though I was living in a district served by a municipal electric utility. Screw the "right to profit", every town should have one of these. [5]RRDtool is for wimps and looks bad too. I had sub-pixel resolution, PNG output and anti-aliased fonts.
Don't confuse simple to use with basic - just because something is easy to operate it doesn't mean that it's incapable of doing some complicated things.
Don't confuse complex with complicated. From m-w.com:
COMPLEX suggests the unavoidable result of a necessary combining and does not imply a fault or failure <a complex recipe>. COMPLICATED applies to what offers great difficulty in understanding, solving, or explaining <complicated legal procedures>.
Many examples spring to mind but the telephone is top of my list. With my phone I can call half way around the world in just a few seconds - heck, even my two year-old nephew can.
The telephone isn't all that simple and yet more basic than you give it credit for. All a telephone does from the user point of view (advanced services aside for the moment) is accept a sequence of numbers that identifies another station somewhere in the world, and attempts to build a bidirectional circuit to that station from available resources. The only reason it appears to be simple is because most people consider phone numbers as very nearly opaque. If the person supplying you with that phone number didn't give you the area code, or you're in a country other than your own and don't know how to get onto the international network, it's not so simple anymore, is it?
But on to the point of my post. Difficulty of use of any piece of equipment is related to two design qualities. First, how many options is a user supplied with? Compare the Macintosh keyboard with the PC keyboard, a mechanical microwave timer with an electronic microwave timer, or a modern PBX station with a Bell System twelve-button POTS phone from the 1970s. A device that offers lots of possibilities right there on the front panel intimidates the inexperienced user and can disorient even the most seasoned. It is possible to offer functionality without disturbing the perception of simplicity by hiding it beneath a trapdoor, as some televisions and VCR's (and TiVo) do.
Many Americans being functionally illiterate, the second quality governing the perceived complexity of the user experience is the amount of reading a user must do to operate the device. Products with thick manuals firmly between the user and the functionality they want are an obvious target, but a more subtle yet influential problem is that some prompts, menu items, dialog boxes, etc. are too hard to (quickly) read. Products that talk too much tend to be perceived as complicated by the uninitiated and annoying by the initiated. Menu items should ideally be no more than one short, ideally monosyllabic, easily recognized word or phrase. Good examples are "Empty Trash", "Clean up", "Quit", "Back". Bad examples are "Empty Recycle Bin" (not so easily recognized, polysyllabic), "Open Web location..." (long, unclear, not so easily recognized: compare to "Go to..."). Menus should place more frequently used options in shallower places. RPN-style "Noun->Verb->Adverb" structures are good, as usually the user knows what they want to manipulate before they know how they want to manipulate it, but consistency is more important than the particular structure.
I am not a trained user experience professional, so take this advice with a salt shaker or two and all your wits.
Palladium is an open architecture (I mean, on paper, that is).
Not if it's patented. Go search on www.uspto.gov for "digital rights management operating system".
It doesnt exisit yet, but the idea is that its not just MS running the show - anyone could be the authority you trust - hell it could be the FSF!
Incorrect. The system as described in Microsoft's patents is based on the premise of transitive trust: BIOS trusts hardware, OS trusts BIOS, application trusts OS therefore application trusts hardware.
One problem is that it's impossible to ship such an OS with a level of trust that preserves competition. If only MSFT is trusted by default, and a scary message must be acknowledged before trusting other parties, most users will use only MSFT software. If only MSFT and people it trusts are trusted by default, and a scary message must be acknowledge before trusting other parties, MSFT gains a lot of power over what people do use (and trust can be centrally revoked, enabling MSFT to partake of a number of slimy business models). If VeriSign or similar is at the root of default trust at the OS level, and a scary message must be acknowledged before trusting other roots, shareware/freeware authors have to pay a tax to VeriSign to create their applications, thus stifling innovation. If no scary message is printed at all, then the point of the whole system is moot.
Anyone can be a trusted source - anyone! This is about hardware enforced trust, not MS literally signing every piece of code that runs on your box.
Have you tried as an individual to get an Authenticode certificate from VeriSign lately? They won't do it because of half-assed reasoning that includes the two meaningless trump words "national security".
If, as you claim, this project is about "hardware enforced trust" then how does a user attempting to insert their own hierarchy of trust distinguish themselves from a virus (or, heaven forbid, a competitor) attempting to insert its own hierarchy of trust?
This is about software trusting hardware and software trusting software. The hardware doesn't need to trust anything, and hardware trusting software is a well-researched and well-practiced problem which requires nothing short of potting whole systems in epoxy to foil attackers. Read Microsoft's patents, not Microsoft's propaganda.
You are correct - this is the same idea as "smart cards" except that its for the masses.
This has nothing to do with the problems smart cards solve. Smart cards attest to the identity of the user, and as people are movable it makes perfect sense for these to be movable as well. Palladium's version of trust has nothing to do with a user proving their identity and only with proving a computer's identity. People don't care about a computer's identity. State-sanctioned spies, content vendors, corporations, software and software vendors do. What does a secure real-time clock do for the average user? Nothing. This is not about solving problems for the end-user.
Releasing the code and a full specification, especially if the code is BSD-licenesed, will prove that MS's intentions and implementations are designed to elevate the entire industry, not just MS.
Incorrect. If there is a patent on loading and identifying a digital rights management operating system its use is governed by Microsoft's licensure of that patent. If systems will (as feared) fail to allow use of the cryptographic processor or potentially even the entire system unless every stage of the boot trusts the next one by signature, that seriously degrades the user serviceability of open-source OSes. If users can set the secure real-time clock then it's clearly not secure. To top it all off, Microsoft is not known for handing out code under terms that allow modification or redistribution, and I fully expect the Palladium source to be released under the same viral "shared-source" look-but-don't-compete license as the CIFS specification and MSDN.
At this point MS could go closed, proprietary, only good for Microsoft, or it could go for open, wide-ranging, available for everyone. It looks like they are learning towards the latter.
History has shown they open things just enough to get maximum traction in any particular campaign. I suspect that, as they have done historically, they will disclose just enough info to allow them some slimy claims about openness and then aggressively leverage those claims to gently or brutally exclude competition on many levels.
This initiative has nothing to do with consumers except to ensure they consume and pay for the privilege.
Any/. readers from California reading this who care to offer a first hand perspective?
Just left California, fwiw...
The shortage was artificial, caused mostly by Enron manipulating the market. I lived in Santa Clara, a city with a municipal electric utility and a municipal generator, and never once experienced a power cut, with minimal changes to my usage patterns, and paid consistently less than PG customers too. But the free market is blameless blah blah.
The co-los in town had a lot of diesel generators in the parking lot, ready to provide some of their own power when the state wanted them off the grid for a little bit. Most large retail facilities in California and a few small ones did dim their interior lighting to save energy, and still do.
His discussion of the legal risks of decrypting these blacklists is fascinating too, and (as he likes to say) "a topic in itself." He would like to open up the source to his SmartFilter-decryption tool but feels the legal risk is too high. How sad is that?
Blocked site lists for filtering software are one of the two classes of copyrighted works explicitly exempt from the DMCA's anti-circumvention provisions. Presumably this also means one can redistribute tools to generate such compilations. IANAL, TINLA.
As a janitor, I refuse to work at a company where employees are allowed to eat in their offices.
Not quite an accurate analogy. As a clean room scrubber, I refuse to work at a company where employees are permitted to eat in the clean room. Or as a plumbing technician, I refuse to work at a company where employees are encouraged to flush everything down the loo and don't know better than to pee all over the floor. I simply refuse to work in places where people are permitted or, worse, encouraged to wallow in their own idiocy and create train wrecks on a daily basis and compel someone else to mop it up for them without the least bit of respect or deference.
In many situations, system administrators are responsible for system uptime and often given zero authority to enforce, create or even suggest policies which get in the way of whiny developers, regardless of the resultant increase in code quality[1]. Talented software engineers are a lot harder to find than talented system administrators because hiring managers perversely ignore most of the people who can do the job right, merely because said applicants are over 35. Most companies would rather try to replace a sysadmin than a software engineer because the chief job of the system administrator in a small-to-midsize organization is to hide and absorb institutional incompetence.
Then again, any software engineer who would demand root on a production system is probably insufficiently skilled to understand basic computing concepts like "separation of privilege" (as seen very recently in OpenSSH), "compartmentalization", "principle of least surprise", and so forth. Far from being engineers in any sense of the term, they're at best "code jockeys" and ought to be physically beaten on a daily basis with classic computer science texts. 90% of them are nothing more than whiners with degrees, and the other 10% design software for the users -- all of them including the poor sot who has to restart that crashy server at 2am every second or third morning.
So, if you can afford to turn down jobs because the software engineers have root access, then hooray for you. But you don't want to get in a pissing contest like that at most companies because the developers will usually win.
I left the technology industry about a year ago, and until more of the antipatterns shake out I don't plan on returning. Unfortunately, the corporate circle jerk has much invested in maintaining these antipatterns so I don't expect the situation will get better soon. As much antipathy as I have for people, professional body piercing sounds like a far preferable career with less bullshit and higher hourly pay. For that matter, so would pizza delivery or auto parts order desk.
Fuck that too. As a system administrator, I refuse to work at a company where all developers have unlimited root access on the production network. I've seen too much stomping about production by developers (and their code) with no sense of Tao, and it's made my life incredibly frustrating in the past. There's no reason for you to be noodling about anywhere near production if the app is well-designed, well-partitioned from the system and keeps its tentacles out of everything.
I've seen those companies that require you to get IT for every little thing. The usual result-- IT cops a major attitude, nothing gets installed, everything breaks, and no one gets a damned thing done.
If your code is a web application, there is no reason, alibi or excuse for your code to run as root, to write files outside of its own chroot jail, to run privileged code, or to bind to privileged ports UNLESS your site uses custom Apache modules or is so big that it must use ASLB. That said, it's nice if a workstation's/usr/local is writable by the user of that workstation and IS leaves a pristine read-only copy around for you or them to rsync if the need arises.
If you develop on Windows, well, there's your problem.
Yes, it's annoying having to track bullshit changes in a huge pile of untested hacks sold into an all-but-captive market, but it shouldn't take much more than 4 hours of hacking perl (or your choice of language) and ImageMagick bindings to do online transcoding the images from TIFF to PNG or GIF or JPEG or even BMP. That will at least get the information and the revenue flowing again, and makes for a usable long-term solution if re-encoding an entire library isn't cost-effective.
When there's something you want to change in your hardware-based rendering, what are you going to do, re-fab the silicon and solder it in?
You can all but program FPGAs in C these days anyway, and a modest stack of FPGAs can do amazing things, fast.
You could start with an architecture similar to Andrew Huang's five-or-so-year-old Tao reconfigurable computing platform, with pipelining de-emphasized. system speed approximately doubled, and (possibly) multi-ported memory added.
But I don't think anyone really ever disagreed with his final point: "The lesson: Publicly traded companies are not the whole computer industry, and the publicly traded stock market is not the whole economy." Was this ever a source of controversy?
This article wasn't published in a trade publication, but in a daily newspaper. Most often the business press wants to hide facts like this from the average Joe, and it's good to see reminders of that in print every now and again so I don't get the urge to fsck myself with another salaried job.
The reason Dell and Gateway and large manufacturers are so important have to do with the support contracts they offer,
White box firms can roll almost instantaneously and often have parts and systems in stock.
the shipping options,
See above.
the warranties,
See above.
the phone support,
Ah, here's a possible failing of the small retailer. The phone support is often relatively weak -- but phone support is pretty much a non-issue when you have minimally sharp desktop people of your own on hand (which you do, if you're a large company).
the willingness and ability to ship next-day in the event of component failure
White box companies can roll almost instantaneously and often have parts and systems in stock.
In short, the security blanket that makes department managers at large companies feel comfortable purchasing those systems.
Corporations are best known for swallowing their own bullshit. It's the same reason COTS software is so prevalent in large organizations, the same reason schmucks pay six and seven figures for crap like Vignette or BroadVision or Dynamo: they want someone they think they can blame, even if they can't.
But [the typical purchasing manager] doesn't know them and here enters the important issue of brand value, identity, and leverage.
Better the devil you do know than the devil that lives entirely in one's mind? It's just another excellent example of the corporation swallowing its own bullshit. I once had a manager describe in hushed tones the Aura of the Brand, of how a brand represents an experience, much like how an infant saying "ma-ma"/* FIXME needs localization */ results in the goddess figure of its life appearing.
Except when it doesn't.
Ladies and gentlemen, we have put the economy and our very lives in the hands of imaginary colossal infants, and THEY NEED SPANKED.
Not to mention that the Dells and Gateways can, in fact, ship in the hundreds of units per day, manufacture in the thousands per week and purchase components in the billions of dollars per year. That's why they're important and has that really ever been a mystery?
And this is important why? This is worth paying extra and getting depersonalized service to who? White-box builders are no less capable of shipping hundreds of barebones systems per day, to order. Dell and Compaq both OEM their finished notebooks from an outfit called Compal. They're not a contract manufacturer, but a turnkey solution for notebook design and manufacturing.
This is what several companies do for the white box market.
This reporter got a good story and then took the wrong angle.
For PHBs and others invested in the worldwide corporate circle-jerk, perhaps. As it is, it's a testament to partial decentralization.
What the large computer companies need to do to stay competitive is find way to cut corners like the smaller companies.
Having worked for a white box retailer not too long ago, I can tell you there isn't that much room for shaving in the large company. Your local white box retailer, especially if the pond is sufficiently small (like Paso Robles, CA), can afford to underpay their help significantly. Dell or Compaq, they's city people, which enlarges their talent pool to burn through but increases the local wage.
Skipping the $300+ dollars a box for M$
Is that what it's up to now? About five years ago, it was $110 for a bundle with all sorts of MSFT and non-MSFT crap^Wsoftware^Wcrap, of which the only piece of marginal utility was the Windoze CD and then only because it had COMMAND.COM on it.
Frankly I wouldn't mind seeing RealPlayer dragged into the street and run over repeatedly by a half-track, but an out-of-date version of that comes free with many OEM bundles too.
No, Intel gave a lot of thought to that. It takes several years to develop a complex CPU like the pentium family.
It only takes a year or two to develop a relatively simple CPU like ARM or MIPS.
RISC designs tend to be far more straightforward and simple. Many computer engineering students implement the MIPS architecture as an exercise. See the Hennessey book (Computer Architecture: A Quantitative Approach, 2nd ed.) to get an idea of just how simple a processor like ARM should be.
Besides, the much-vaunted new feature of the PPro was the CISC->RISC translator, and it shouldn't take much to rejig that to handle 16-bit mode more effectively if the market (asses that they are) demands it.
Y'all oughta fire everyone but Taco and maybe Hemos so we won't have to see blatant repeats like this, especially when there's real news being REJECTED left and right. This s(h)ite is starting to piss me off.
The tool from MS has marketspeak in place of information except for "We decided to support V4, we didn't bother to retarget our compilers to V5 for our traitorous former buddies at Intel, and if XScale's V4 compat is weak, that's their problem." It is, in fact, but that doesn't make MSFT's laziness any less lame.
On the other hand, Intel often gives little thought to enhancing performance of old code on new processors. If memory serves me right, Intel's Pentium Pro ran 16-bit code embarrassingly slowly.
We have also encountered Web sites of issue advocacy groups that have positioned the audio link to an NPR story such that one cannot tell that NPR is not supporting their cause.
You are lying. There is no way that one could link to a stream of a fair and impartial newscast (links to streams must be to the whole stream, from beginning to end, remember)
You are ignorant. RealMedia servers will accept parameters such as "start=hh:mm:ss:f" (and possibly a matching "length" parameter) in the query string of a request. You can create a ram file that shuffles the news however you want it, and further insert your own clips with sound-alikes. So, my little politician, it behooves you to learn what's possible and what's out there instead of making these idiotic little assumptions to score points on your high school debate team.
can't it query the current monitor like Macs and Winders do (DPMS)?
Why, yes, it can. X requires a list of useful resolutions, yes, but it can ask the monitor for the ModeLine info and negotiate a mode compatible with the card and the monitor automatically. Most automated X setup programs put the "usual" resolutions in the XF86Config for you and X selects the best available when it starts.
I only ever change my resolution -- ahem, viewport size -- to watch too-small pr0n videos, anyway.
Slashdot Mirror
-jhp
-jhp
-jhp
-jhp
The more I read /. the more I find Wallace's misanthropy rubbing off on me.
-jhp
That said, a 33MHz machine might not have the UARTs and the speed to handle 115.2kbps data, which is the 1-Wire high-speed rate.
-jhp
Yes, I agree it was grossly overcomplicated, and if I were doing the same thing today I'd probably have had the TINI post data to the app server instead, so as to cut the perl script out of the loop. But I didn't really trust the TINI as much more than a really smart 1-Wire interface, so in reality I'd probably design with the RS232-based 1-Wire interface card instead, or use a PIC to do all this. :-)
-jhp
[1] It was a bar fridge and the freezer frequently affected the thermostat due to their closeness. In line with programmer virtue #1, I wanted to wait to defrost until the temperature in the fridge put my food at risk.
[2] This was my first experience with surface-mount soldering. The new rev of the sockets board has a lot less cool stuff on it now -- the LCD interface is gone, for one.
[3]The TINI's Java environment wasn't hefty enough to handle PostgreSQL JDBC drivers so something else had to do this.
[4]It was 2001. I was trying to be nice to the electric grid and my own power bill, even though I was living in a district served by a municipal electric utility. Screw the "right to profit", every town should have one of these.
[5]RRDtool is for wimps and looks bad too. I had sub-pixel resolution, PNG output and anti-aliased fonts.
But on to the point of my post. Difficulty of use of any piece of equipment is related to two design qualities. First, how many options is a user supplied with? Compare the Macintosh keyboard with the PC keyboard, a mechanical microwave timer with an electronic microwave timer, or a modern PBX station with a Bell System twelve-button POTS phone from the 1970s. A device that offers lots of possibilities right there on the front panel intimidates the inexperienced user and can disorient even the most seasoned. It is possible to offer functionality without disturbing the perception of simplicity by hiding it beneath a trapdoor, as some televisions and VCR's (and TiVo) do.
Many Americans being functionally illiterate, the second quality governing the perceived complexity of the user experience is the amount of reading a user must do to operate the device. Products with thick manuals firmly between the user and the functionality they want are an obvious target, but a more subtle yet influential problem is that some prompts, menu items, dialog boxes, etc. are too hard to (quickly) read. Products that talk too much tend to be perceived as complicated by the uninitiated and annoying by the initiated. Menu items should ideally be no more than one short, ideally monosyllabic, easily recognized word or phrase. Good examples are "Empty Trash", "Clean up", "Quit", "Back". Bad examples are "Empty Recycle Bin" (not so easily recognized, polysyllabic), "Open Web location..." (long, unclear, not so easily recognized: compare to "Go to..."). Menus should place more frequently used options in shallower places. RPN-style "Noun->Verb->Adverb" structures are good, as usually the user knows what they want to manipulate before they know how they want to manipulate it, but consistency is more important than the particular structure.
I am not a trained user experience professional, so take this advice with a salt shaker or two and all your wits.
-jhp
-jhp
One problem is that it's impossible to ship such an OS with a level of trust that preserves competition. If only MSFT is trusted by default, and a scary message must be acknowledged before trusting other parties, most users will use only MSFT software. If only MSFT and people it trusts are trusted by default, and a scary message must be acknowledge before trusting other parties, MSFT gains a lot of power over what people do use (and trust can be centrally revoked, enabling MSFT to partake of a number of slimy business models). If VeriSign or similar is at the root of default trust at the OS level, and a scary message must be acknowledged before trusting other roots, shareware/freeware authors have to pay a tax to VeriSign to create their applications, thus stifling innovation. If no scary message is printed at all, then the point of the whole system is moot.
Have you tried as an individual to get an Authenticode certificate from VeriSign lately? They won't do it because of half-assed reasoning that includes the two meaningless trump words "national security". If, as you claim, this project is about "hardware enforced trust" then how does a user attempting to insert their own hierarchy of trust distinguish themselves from a virus (or, heaven forbid, a competitor) attempting to insert its own hierarchy of trust?This is about software trusting hardware and software trusting software. The hardware doesn't need to trust anything, and hardware trusting software is a well-researched and well-practiced problem which requires nothing short of potting whole systems in epoxy to foil attackers. Read Microsoft's patents, not Microsoft's propaganda.
This has nothing to do with the problems smart cards solve. Smart cards attest to the identity of the user, and as people are movable it makes perfect sense for these to be movable as well. Palladium's version of trust has nothing to do with a user proving their identity and only with proving a computer's identity. People don't care about a computer's identity. State-sanctioned spies, content vendors, corporations, software and software vendors do. What does a secure real-time clock do for the average user? Nothing. This is not about solving problems for the end-user. Incorrect. If there is a patent on loading and identifying a digital rights management operating system its use is governed by Microsoft's licensure of that patent. If systems will (as feared) fail to allow use of the cryptographic processor or potentially even the entire system unless every stage of the boot trusts the next one by signature, that seriously degrades the user serviceability of open-source OSes. If users can set the secure real-time clock then it's clearly not secure. To top it all off, Microsoft is not known for handing out code under terms that allow modification or redistribution, and I fully expect the Palladium source to be released under the same viral "shared-source" look-but-don't-compete license as the CIFS specification and MSDN. History has shown they open things just enough to get maximum traction in any particular campaign. I suspect that, as they have done historically, they will disclose just enough info to allow them some slimy claims about openness and then aggressively leverage those claims to gently or brutally exclude competition on many levels.This initiative has nothing to do with consumers except to ensure they consume and pay for the privilege.
-jhp
The shortage was artificial, caused mostly by Enron manipulating the market. I lived in Santa Clara, a city with a municipal electric utility and a municipal generator, and never once experienced a power cut, with minimal changes to my usage patterns, and paid consistently less than PG customers too. But the free market is blameless blah blah.
The co-los in town had a lot of diesel generators in the parking lot, ready to provide some of their own power when the state wanted them off the grid for a little bit. Most large retail facilities in California and a few small ones did dim their interior lighting to save energy, and still do.
-jhp
-jhp
-jhp
In many situations, system administrators are responsible for system uptime and often given zero authority to enforce, create or even suggest policies which get in the way of whiny developers, regardless of the resultant increase in code quality[1]. Talented software engineers are a lot harder to find than talented system administrators because hiring managers perversely ignore most of the people who can do the job right, merely because said applicants are over 35. Most companies would rather try to replace a sysadmin than a software engineer because the chief job of the system administrator in a small-to-midsize organization is to hide and absorb institutional incompetence.
Then again, any software engineer who would demand root on a production system is probably insufficiently skilled to understand basic computing concepts like "separation of privilege" (as seen very recently in OpenSSH), "compartmentalization", "principle of least surprise", and so forth. Far from being engineers in any sense of the term, they're at best "code jockeys" and ought to be physically beaten on a daily basis with classic computer science texts. 90% of them are nothing more than whiners with degrees, and the other 10% design software for the users -- all of them including the poor sot who has to restart that crashy server at 2am every second or third morning.
I left the technology industry about a year ago, and until more of the antipatterns shake out I don't plan on returning. Unfortunately, the corporate circle jerk has much invested in maintaining these antipatterns so I don't expect the situation will get better soon. As much antipathy as I have for people, professional body piercing sounds like a far preferable career with less bullshit and higher hourly pay. For that matter, so would pizza delivery or auto parts order desk.-jhp
-jhp
If you develop on Windows, well, there's your problem.
-jhp
-jhp
You could start with an architecture similar to Andrew Huang's five-or-so-year-old Tao reconfigurable computing platform, with pipelining de-emphasized. system speed approximately doubled, and (possibly) multi-ported memory added.
-jhp
Except when it doesn't.
Ladies and gentlemen, we have put the economy and our very lives in the hands of imaginary colossal infants, and THEY NEED SPANKED.
And this is important why? This is worth paying extra and getting depersonalized service to who? White-box builders are no less capable of shipping hundreds of barebones systems per day, to order. Dell and Compaq both OEM their finished notebooks from an outfit called Compal. They're not a contract manufacturer, but a turnkey solution for notebook design and manufacturing.This is what several companies do for the white box market.
For PHBs and others invested in the worldwide corporate circle-jerk, perhaps. As it is, it's a testament to partial decentralization.-jhp
Frankly I wouldn't mind seeing RealPlayer dragged into the street and run over repeatedly by a half-track, but an out-of-date version of that comes free with many OEM bundles too.
-jhp
Besides, the much-vaunted new feature of the PPro was the CISC->RISC translator, and it shouldn't take much to rejig that to handle 16-bit mode more effectively if the market (asses that they are) demands it.
-jhp
Y'all oughta fire everyone but Taco and maybe Hemos so we won't have to see blatant repeats like this, especially when there's real news being REJECTED left and right. This s(h)ite is starting to piss me off.
-jhp
On the other hand, Intel often gives little thought to enhancing performance of old code on new processors. If memory serves me right, Intel's Pentium Pro ran 16-bit code embarrassingly slowly.
-jhp
-jhp
I only ever change my resolution -- ahem, viewport size -- to watch too-small pr0n videos, anyway.
-jhp