As much as I like ThinkGeek, their selection is limited to gadgets. I found that assembling and -- to my parents dismay -- disassembling things are what really grabbed my interest.
I would take a look at the various kits from American Science & Surplus. There are a number of other sites (e.g., Carl's Electronics) which have even more kits, but I haven't ordered from them so I can't say whether they're worthwhile or not. (These days, most of my toys come from DigiKey, and not in kit form.)
Based on our internal BlackBerry users mailing list at work, it seems the trackball is usually the first to break. The number of e-mails asking if it's possible to "check e-mail without scrolling {left,right,up,down}" swamps all other topics. The DIY cleaning tips available out there are hit-and-miss. I've been lucky; the two times my 8820 has stopped tracking, the DIY tips worked.
If the iPhone were allowed into our internal network, I would probably migrate over in a heartbeat. However, this is one area where RIM has everyone else beat: they know how to sell to the corporate IT folks. My BB -- which I paid for, personally -- is controlled by our friendly BOFHs and has their policies applied, like locking the screen after 10 minutes of inactivity and requiring me to enter a password. They can also remotely wipe my BB at will. It's these kinds of features which give them warm fuzzies.
Just to point out, if you RTFP (post) mattytee doesn't say it's ok, he says it's "hinky." Which might NOT mean okay. I admit, I don't know what it ACTUALLY means, so it might mean "good."
X was never that great at running apps over general networks. The protocol requires so many round-trip queries that it's highly dependent on latency -- e.g., it's ok for getting an xterm on the server two buildings over, but painful once you start measuring the distance in miles. Although many aspects of computing devices are getting faster, the speed of light has held steady.
VNC was an interesting approach to the problem. All of the initial queries are up-front, then it's just the two sides pushing events at each other (mostly "draw this pixmap" and "move the cursor"). I say "interesting" because I would not have expected remote framebuffers to be practical (I was still on dial-up when I first started using it).
NX is closer to what I had envisioned. However, getting it to work in the enterprise has been a challenge, at least in my experience. I can compile TightVNC for RHEL 3, for instance, with little more that a configure, make, make install. FreeNX, OTOH, not so friendly. (And installing personally licensed commercial software, like NoMachine's implementation, is a big no-no here.)
Assuming Sprint's press release is accurate (which, based on previous juvenile behavior by Cogent, I'm inclined to believe them), Cogent has effectively cut themselves off from the Internet.
Put simply, they've put their AS onto the net without having the necessary routing arrangements in place. They are, at best, a kinda-sorta-maybe-if-you're-lucky part of the Internet right now. They want to be a transit provider without doing the messy stuff like getting contracts in place. Good luck with that.
This is why I should never be a manager (or, by extension, evil overlord). I'll make some typically wise-ass remark ("Gee, why don't you just go drop a few random tables from the customer database?"), only to have one of my minions dutifully carry this out.
cloud computing is nothing more than managed colocation services. or sophisticated hosting services..
Er... sort of, but not really. In a sophisticated, managed hosting environment, I'd want tighter SLAs on the hardware than in a cloud computing environment (at least, given what I've worked with on EC2). I'd be willing to put a database on the former; on the latter, only rendering machines (where "render" can mean rendering web pages, movies, the billionth digit of pi, etc.).
The "cloud" part of "cloud computing" comes from how you would draw it on a network or system architecture diagram. Clouds are used when you don't care about the details. For example, getting two servers to talk to each other across the Internet, the Internet is a cloud. You don't care if the network connections involved are DSL, cable, ATM, IP-over-avian carriers, etc.; all you care about is that packets sent into the cloud magically arrive at the other end.
Likewise, with cloud computing, you don't care about the details of the server (Opteron/Intel, SCSI/SATA, 16 GB or 32 GB, etc.); all you care about is that you send a task into this cloud and it spits an answer back at you. If the underlying architecture does matter to you, you shouldn't be using cloud computing!
I'm guessing this was scrapped mainly because it's not an easy story to make it interesting for the mainstream audience. It's one thing to show engines on fire, chemicals blowing up, and people getting zapped by high voltage; quite another to show Grant fiddling with a hex editor.
Their previous security exploits (for example, hijacking the fingerprint reader) were still very tangible hacks (wax impressions, photocopied prints, etc.). I'm guessing 99.9% of/. readers thinks this is worthwhile and 99.9% watch Mythbusters; however, this does not comprise 99.9% of Mythbuster's audience (or even a majority, I'd bet).
Oh, they've made a bunch of lower-profile mistakes with the company I work for. We have about 50k RHEL installs across our sites; actual issues with their image (especially the kernel "improvements" in RHEL3) that affect production servers are routinely ignored or dismissed.
We've considered dumping them, but this is the main supported Oracle platform -- and we're too invested in Oracle to switch away. I realize the plural of anecdote is not data, but our experience is not unique.
They gave away Microsoft's private keys to someone who called them
Not quite. Microsoft's private key wasn't compromised; their identity was stolen. The attacker convinced VeriSign to sign his certificate claiming to be "Microsoft Corporation." The whole point of PKI is to never transmit your private key, even to an authority like VeriSign. As usual, the technology is secure; it's the people running it who aren't.
ssh is vulnerable unless you set up the fingerprint of the host you're talking to beforehand.
How many times have you just ignored this warning and allowed ssh to continue? Warning: Permanently added 'foo.bar.com,23.227.17.89' (RSA) to the list of known hosts.
If you're allowing that, it's trivial for someone to perform a man-in-the-middle attack on your connection. You have no idea if you've accepted the host's actual key or the hijacker's key.
WTF? My team uses Fortify to analyze our Java webapps (compiled on the Sun JDK and running on their JRE), which is then deployed to Linux servers running RHEL 5. HTTP connectivity for the apps is provided by Jetty; the apps themselves connect to Oracle databases (using C3P0 for connection pooling).
With Fortify 4.0, I griped that it provided no value that we didn't already get with FindBugs (for free). The 5.0 release (along with the workbench, which provides better information than the HTML report), however, did catch a few bugs which weren't caught by FindBugs. We now run both tools in our automated Hudson builds.
Where, exactly, are the Microsoft products in the above list?
Google is being of limited help here. The main link I'm finding is to Liquidmetal Technologies, which is producing Liquidmetal and Vitreloy -- zirconium-based alloys which are amorphous in structure (hence the "liquid" in the name) but are otherwise solid in appearance and use (and much stronger than stainless steel or titanium). This is not something one would be pumping through heat tubes to cool a CPU.
Obviously, mercury is out due to its toxicity. My initial thought was they're using metal bits in a suspension, but I have doubts as to whether this would actually do anything useful. Deeper searching yields this page, which describes a gallium/indium/tin alloy which is liquid at room temperature. Wikipedia'a entry for gallium concurs, saying, "It has been suggested that a liquid gallium-tin alloy could be used to cool computer chips in place of water."
If you mean any modern Linux distro I would say that it is better than W2K. Lets start from the top...
Solid kernel.
Solid GUI base (X)
Solid GUI (take your pick, XFCE, GNOME, KDE, etc)
This is a bunch of FUD which isn't based on any facts that I'm aware of. The NT/2k kernel is a fairly solid piece of work. It does feel a bit overengineered in some respects, but I'd like to see some evidence that it's not solid.
That said, the userland DLLs and APIs which surround it are a mismash of inconsistencies and ill-thought out abstractions.
As for GUI bases, the Windows assumption that there's always a display with set policies (window station, window manager, win32k.sys, etc.) is unfortunate for servers, but a definite improvement over X for desktops. Sure, X abstracts out the transport layer allowing for decoupling of the application server and display server, but nobody uses this anymore -- both the protocol (too many round-trip queries kill performance over anything but the lowest latency connections) and limitations (no sound, printing is an after thought, changing configuration midsession) were its death. (We use VNC or NX for serious work over remote X sessions.)
And aren't "solid GUI" and "take your pick" contradictory? If the GUI was solid, why can't I code against a single API and have it "just work?" The fact that a user can tell that a given program was coded against a different toolkit might have been amusing in the late 80s and early 90s.
Don't get me wrong -- I love Linux as a coding platform and for deploying a fleet of thousands of web/application servers, all alike. The kernel, though, I'd rate as equal to NT (especially given the latency and driver integration issues we've had, notably with HP's recent servers). The desktop experience, though, is decidedly inferior.
I had hopes for Berlin/Fresco and other from-the-ground-up rewrites for awhile -- my opinion on X is that it's a lost cause; we must "nuke it from orbit. It's is the only way to be sure." Alas, they all seem to have stalled.
Nah, it goes deeper than that. Dealing with Sun as a target platform is frustrating as can be. Are they investing money on SPARC or is it dead? Do we need to verify our app run on Solaris 10 or are they dropping it for Linux? Should we develop our app for an old release of Java, or do we require our clients to have the latest version (including the OS patches required to install it?).
In fact, garbage collection can be faster than malloc/free, and has been demonstrated to be so in some cases. (One obvious case where a collector outperforms manual memory management is when the collector never has to run.) That's an unfair comparison. I'd be more impressed in the latter case if it was still faster than just malloc() without free().:-)
I think I can see where this may have made sense at some level. But, like all good ideas, this probably got corrupted in a committee.
First, forget that you're a geek. You just need a computer for the basics: e-mailing friends and colleagues, writing papers for school/work, a presentation every other month or so, and maybe a spreadsheet for crunching numbers for the boss (or keeping tabs on your CD collection -- your geek friends keep telling you about how this is a horrible misuse and you should put together a database, but, hey, it works). Oh, and a game or two: FreeCell and Sudoku.
I have a number of friends and relatives who fit this profile. More often than not, they can't even keep this working as expected: their systems are littered with every virus known to the virus checkers out there and then some; the free version of Norton on their system expired years ago (but is still happily hogging resources, too). The system came with Windows 98, but a friend "who's good with computers" tossed a copy of XP Home from his system on there. Those WGA messages are real, which also means security updates haven't been applied since the first season of American Idol.
Many of these folks would gladly shell out $25/month if it meant that their computer "just worked." They don't care that they don't "own" the rest of the software on their machine as long as they can get their work done.
The vendor -- in this case, Microsoft -- sees an appealing monetary model here. That's $300/year per customer. Compared to the bargain rates they charge HP, Dell, etc., for XP and Office (say, $100-$200?) and a three-year lifespan of a system, this is a huge cash cow.
However, they're forgetting something: the system has to "just work" if the customers are going to buy this. Without severely tightening down what can be done to the OS (and risking likely anti-trust issues with that), this will involve a lot of tech support to get it working. The more likely alternative: frustrated customers who don't see any reason to continue paying what really amounts to a monthly tax for the same old broken computer.
Slashdot Mirror
As much as I like ThinkGeek, their selection is limited to gadgets. I found that assembling and -- to my parents dismay -- disassembling things are what really grabbed my interest.
I would take a look at the various kits from American Science & Surplus. There are a number of other sites (e.g., Carl's Electronics) which have even more kits, but I haven't ordered from them so I can't say whether they're worthwhile or not. (These days, most of my toys come from DigiKey, and not in kit form.)
Based on our internal BlackBerry users mailing list at work, it seems the trackball is usually the first to break. The number of e-mails asking if it's possible to "check e-mail without scrolling {left,right,up,down}" swamps all other topics. The DIY cleaning tips available out there are hit-and-miss. I've been lucky; the two times my 8820 has stopped tracking, the DIY tips worked.
If the iPhone were allowed into our internal network, I would probably migrate over in a heartbeat. However, this is one area where RIM has everyone else beat: they know how to sell to the corporate IT folks. My BB -- which I paid for, personally -- is controlled by our friendly BOFHs and has their policies applied, like locking the screen after 10 minutes of inactivity and requiring me to enter a password. They can also remotely wipe my BB at will. It's these kinds of features which give them warm fuzzies.
Just to point out, if you RTFP (post) mattytee doesn't say it's ok, he says it's "hinky." Which might NOT mean okay. I admit, I don't know what it ACTUALLY means, so it might mean "good."
It could be a typo for "kinky." Which, I can only imagine, would be included in the comprehensive list of fetishes.
X was never that great at running apps over general networks. The protocol requires so many round-trip queries that it's highly dependent on latency -- e.g., it's ok for getting an xterm on the server two buildings over, but painful once you start measuring the distance in miles. Although many aspects of computing devices are getting faster, the speed of light has held steady.
VNC was an interesting approach to the problem. All of the initial queries are up-front, then it's just the two sides pushing events at each other (mostly "draw this pixmap" and "move the cursor"). I say "interesting" because I would not have expected remote framebuffers to be practical (I was still on dial-up when I first started using it).
NX is closer to what I had envisioned. However, getting it to work in the enterprise has been a challenge, at least in my experience. I can compile TightVNC for RHEL 3, for instance, with little more that a configure, make, make install. FreeNX, OTOH, not so friendly. (And installing personally licensed commercial software, like NoMachine's implementation, is a big no-no here.)
Assuming Sprint's press release is accurate (which, based on previous juvenile behavior by Cogent, I'm inclined to believe them), Cogent has effectively cut themselves off from the Internet.
Put simply, they've put their AS onto the net without having the necessary routing arrangements in place. They are, at best, a kinda-sorta-maybe-if-you're-lucky part of the Internet right now. They want to be a transit provider without doing the messy stuff like getting contracts in place. Good luck with that.
Insightful? Insightful? I was being flippant. :-P
This is why I should never be a manager (or, by extension, evil overlord). I'll make some typically wise-ass remark ("Gee, why don't you just go drop a few random tables from the customer database?"), only to have one of my minions dutifully carry this out.
What if the uncooperative human is the one *controlling* the robots?
cloud computing is nothing more than managed colocation services. or sophisticated hosting services..
Er... sort of, but not really. In a sophisticated, managed hosting environment, I'd want tighter SLAs on the hardware than in a cloud computing environment (at least, given what I've worked with on EC2). I'd be willing to put a database on the former; on the latter, only rendering machines (where "render" can mean rendering web pages, movies, the billionth digit of pi, etc.).
The "cloud" part of "cloud computing" comes from how you would draw it on a network or system architecture diagram. Clouds are used when you don't care about the details. For example, getting two servers to talk to each other across the Internet, the Internet is a cloud. You don't care if the network connections involved are DSL, cable, ATM, IP-over-avian carriers, etc.; all you care about is that packets sent into the cloud magically arrive at the other end.
Likewise, with cloud computing, you don't care about the details of the server (Opteron/Intel, SCSI/SATA, 16 GB or 32 GB, etc.); all you care about is that you send a task into this cloud and it spits an answer back at you. If the underlying architecture does matter to you, you shouldn't be using cloud computing!
Since it's the financial industry you probably won't be working there long :P
Well, someone has to be there to shut off the servers and sell them for scrap on eBay...
I'm guessing this was scrapped mainly because it's not an easy story to make it interesting for the mainstream audience. It's one thing to show engines on fire, chemicals blowing up, and people getting zapped by high voltage; quite another to show Grant fiddling with a hex editor.
Their previous security exploits (for example, hijacking the fingerprint reader) were still very tangible hacks (wax impressions, photocopied prints, etc.). I'm guessing 99.9% of /. readers thinks this is worthwhile and 99.9% watch Mythbusters; however, this does not comprise 99.9% of Mythbuster's audience (or even a majority, I'd bet).
Oh, they've made a bunch of lower-profile mistakes with the company I work for. We have about 50k RHEL installs across our sites; actual issues with their image (especially the kernel "improvements" in RHEL3) that affect production servers are routinely ignored or dismissed.
We've considered dumping them, but this is the main supported Oracle platform -- and we're too invested in Oracle to switch away. I realize the plural of anecdote is not data, but our experience is not unique.
The entire mess could be avoided if taxes were collected based on the seller's location, not the buyer's.
This doesn't quite work, either. If this happened, every company would relocate to Oregon or another state which does not charge sales tax.
Well, ok... it "works" just fine for me :-), but it would essentially destroy sales tax revenue for all catalog/internet purchases.
They gave away Microsoft's private keys to someone who called them
Not quite. Microsoft's private key wasn't compromised; their identity was stolen. The attacker convinced VeriSign to sign his certificate claiming to be "Microsoft Corporation." The whole point of PKI is to never transmit your private key, even to an authority like VeriSign. As usual, the technology is secure; it's the people running it who aren't.
In Soviet Russia, phone locates you!
ssh is vulnerable unless you set up the fingerprint of the host you're talking to beforehand.
How many times have you just ignored this warning and allowed ssh to continue?
Warning: Permanently added 'foo.bar.com,23.227.17.89' (RSA) to the list of known hosts.
If you're allowing that, it's trivial for someone to perform a man-in-the-middle attack on your connection. You have no idea if you've accepted the host's actual key or the hijacker's key.
WTF? My team uses Fortify to analyze our Java webapps (compiled on the Sun JDK and running on their JRE), which is then deployed to Linux servers running RHEL 5. HTTP connectivity for the apps is provided by Jetty; the apps themselves connect to Oracle databases (using C3P0 for connection pooling).
With Fortify 4.0, I griped that it provided no value that we didn't already get with FindBugs (for free). The 5.0 release (along with the workbench, which provides better information than the HTML report), however, did catch a few bugs which weren't caught by FindBugs. We now run both tools in our automated Hudson builds.
Where, exactly, are the Microsoft products in the above list?
Google is being of limited help here. The main link I'm finding is to Liquidmetal Technologies, which is producing Liquidmetal and Vitreloy -- zirconium-based alloys which are amorphous in structure (hence the "liquid" in the name) but are otherwise solid in appearance and use (and much stronger than stainless steel or titanium). This is not something one would be pumping through heat tubes to cool a CPU.
Obviously, mercury is out due to its toxicity. My initial thought was they're using metal bits in a suspension, but I have doubts as to whether this would actually do anything useful. Deeper searching yields this page, which describes a gallium/indium/tin alloy which is liquid at room temperature. Wikipedia'a entry for gallium concurs, saying, "It has been suggested that a liquid gallium-tin alloy could be used to cool computer chips in place of water."
Any materials experts out there care to comment?
If you mean any modern Linux distro I would say that it is better than W2K. Lets start from the top...
This is a bunch of FUD which isn't based on any facts that I'm aware of. The NT/2k kernel is a fairly solid piece of work. It does feel a bit overengineered in some respects, but I'd like to see some evidence that it's not solid.
That said, the userland DLLs and APIs which surround it are a mismash of inconsistencies and ill-thought out abstractions.
As for GUI bases, the Windows assumption that there's always a display with set policies (window station, window manager, win32k.sys, etc.) is unfortunate for servers, but a definite improvement over X for desktops. Sure, X abstracts out the transport layer allowing for decoupling of the application server and display server, but nobody uses this anymore -- both the protocol (too many round-trip queries kill performance over anything but the lowest latency connections) and limitations (no sound, printing is an after thought, changing configuration midsession) were its death. (We use VNC or NX for serious work over remote X sessions.)
And aren't "solid GUI" and "take your pick" contradictory? If the GUI was solid, why can't I code against a single API and have it "just work?" The fact that a user can tell that a given program was coded against a different toolkit might have been amusing in the late 80s and early 90s.
Don't get me wrong -- I love Linux as a coding platform and for deploying a fleet of thousands of web/application servers, all alike. The kernel, though, I'd rate as equal to NT (especially given the latency and driver integration issues we've had, notably with HP's recent servers). The desktop experience, though, is decidedly inferior.
I had hopes for Berlin/Fresco and other from-the-ground-up rewrites for awhile -- my opinion on X is that it's a lost cause; we must "nuke it from orbit. It's is the only way to be sure." Alas, they all seem to have stalled.
Great! I'll just resolve ftp.internic.net from the address I have on file for the L-root server...
Nah, it goes deeper than that. Dealing with Sun as a target platform is frustrating as can be. Are they investing money on SPARC or is it dead? Do we need to verify our app run on Solaris 10 or are they dropping it for Linux? Should we develop our app for an old release of Java, or do we require our clients to have the latest version (including the OS patches required to install it?).
How about billing prisoners for their incarceration or torture? Apparently it's not just a fantasy from Brazil .
Don't fight it son, confess quickly. If you hold out too long, you could jeopardize your credit rating.
Anyone want to help fund an ad for OpenOffice on MS Works?
I think I can see where this may have made sense at some level. But, like all good ideas, this probably got corrupted in a committee.
First, forget that you're a geek. You just need a computer for the basics: e-mailing friends and colleagues, writing papers for school/work, a presentation every other month or so, and maybe a spreadsheet for crunching numbers for the boss (or keeping tabs on your CD collection -- your geek friends keep telling you about how this is a horrible misuse and you should put together a database, but, hey, it works). Oh, and a game or two: FreeCell and Sudoku.
I have a number of friends and relatives who fit this profile. More often than not, they can't even keep this working as expected: their systems are littered with every virus known to the virus checkers out there and then some; the free version of Norton on their system expired years ago (but is still happily hogging resources, too). The system came with Windows 98, but a friend "who's good with computers" tossed a copy of XP Home from his system on there. Those WGA messages are real, which also means security updates haven't been applied since the first season of American Idol.
Many of these folks would gladly shell out $25/month if it meant that their computer "just worked." They don't care that they don't "own" the rest of the software on their machine as long as they can get their work done.
The vendor -- in this case, Microsoft -- sees an appealing monetary model here. That's $300/year per customer. Compared to the bargain rates they charge HP, Dell, etc., for XP and Office (say, $100-$200?) and a three-year lifespan of a system, this is a huge cash cow.
However, they're forgetting something: the system has to "just work" if the customers are going to buy this. Without severely tightening down what can be done to the OS (and risking likely anti-trust issues with that), this will involve a lot of tech support to get it working. The more likely alternative: frustrated customers who don't see any reason to continue paying what really amounts to a monthly tax for the same old broken computer.