How much support does this provide for accessing hard drives? Autodetection? Automounting? UMSDOS? Detecting Config files on the drive? Rescuing broken configurations?
Sure, there are applications for which total portability and a USB key are fine. But often I'd like to be able to work with files on a standalone machine, access downloaded material, etc., even though the underlying machine is configured for Windows. It'd be especially nice to have features like recognizing a/home directory or a/var spool.
Another environment for which this is useful is little server machines hanging off an Internet connection where you're not monitoring security all the time. The thing can sit around serving web pages, running intrusion detection, acting as a honeypot or teergrube for annoying spammers, etc., and if it gets attacked and rooted, you just reboot. Perhaps a more security-oriented distro would be better for this, since this sounds more desktop-oriented, but it'd be nice to have one consistent environment.
It's kind of like Seattle without Microsoft or Boeing... Back in the 70s, when the government was playing lottery games deciding which of its citizens to enslave, and young men had to evaluate their options for how to deal with that, the only parts of Canada I'd been to were The Frozen North, aka Toronto and Montreal and Sudbury. If I'd seen Vancouver, that choice would have seemed more attractive. As it was, I didn't draw a low number, didn't win the green suit or the all-expenses-paid vacation in exciting tropical Southeast Asia, and haven't bought another ticket from those scammers since.
But apparently the Bush Administration is looking for new recruits for the Draft Board. They haven't said when the Homeland Security Patriotic Anti-Terrorist Force starts. Yet.
Re:Unix Tab-Separated ASCII Files vs. XML
on
Effective XML
·
· Score: 1
I'm not arguing that XML is useless or not powerful - it's possible to do really good standardization if you want to. SQL for relational databases is also pretty good at standardizability and readability (so I'll dispute your "unheard of" assertion, and point out that you're using the term "schema"....) I am a bit concerned that people will be too likely to reimplement a lot of the 1970s hierarchical database structure things that we replaced with relational databases during the late 1980s-1990s, though XML probably makes it easier to do that well as opposed to badly.
But just as you can write undecipherable undocumented bad spaghetti code in almost any computer or non-computer language, you can also write uselessly non-standardizable XML that nobody can communicate with independently, even if you _aren't_ deliberately trying to obfuscate it. Microsoft Office's latest version uses XML for data interchange between components, and I've heard people assert that it's not possible for a programmer who isn't buying a bunch of APIs (or otherwise getting Office to do tasks) to communicate with it dependably. I haven't verified that myself, but I've seen the HTML code produced by previous versions, so I'm inclined to believe it.
Unix Tab-Separated ASCII Files vs. XML
on
Effective XML
·
· Score: 4, Interesting
Sure, XML isn't inherently that deep - but neither are the tab-separated ASCII files which Unix tools used to do all kinds of really powerful things. Similarly, LISP property lists aren't that complex. XML's a bit more flexible, and carries enough decoration with it that people are willing to use it for building interfaces that they might not build using ASCII or XDR. And anything that lets the EDI people replace their stuff with simpler, more open technology is good too..
Yup. They're probably pretty customizable - the articles tend to price them as $50-75K for non-mass-produced, and you could probably add an extra 20 kilos of cargo capacity without overly trashing the battery life. I'm not sure the current wholesale prices for cocaine, but if it's $10/gram for delivered to the US (as opposed to $1/gram in Colombia), you could about pay for the sub on its first trip, and definitely PROFIT on the later trips.
On the other hand, during the 80s disco years, the profit on importing coke was in the $1-10M per Cessna-load range, so you could afford to lose half your shipments, and there were plenty of pilots willing to risk a few years in jail for a share of the profits if they win. The profit margin's probably not still 90%, but that's mostly because it's so easy for the importers; even at 50% you can lose almost half your shipments, so as long as you've got enough capital to risk losing a couple of containerloads before one gets through, your real problem becomes making reliable connections with distributors and not getting ripped off on delivery.
There are a lot of attacks against anonymity going on - this section in the spam bill, attempts to require whois data to contain your True Name, ICBM address, subpoena address, etc., the Post Office's requirements that not only their customers but also their competitor's customers provide True Name and True Address when renting mailboxes, etc. These requirements don't only force self-incrimination from real criminals, but from everybody. Sure, putting it in a spam bill theoretically only applies if you're spamming, but it's just another brick in the wall.
The ban on "misleading" headers is unclearly worded, but appears to ban sending email without headers that are actively helpful to "law enforcement". It's not clear whether this bans anonymous remailers in the US, because some of the mail they remove headers from _could_ be from spammers, even though spam is an abuse of what they're intended for. It does look like the bureaucrats (mainly FTC) who implement the policies will have enough definition-making ability that they could probably treat it that way.
In general, the bill fails to differentiate between forging a From or contact email address that's a real email address belonging to someone else (whether joe-job or random), forging a "nobody"-type address at a real domain belonging to someone else, forging a fake address at a real domain belonging to someone else, using a fake address at a real domain that the spammer's allowed to use, using an address at a fake domain, or other things. Forging the address of a real person is highly annoying to them, and might already legally count as forgery. Forging yet another bogus hotmail address is annoying to hotmail, but less annoying than a real address; using "dont-bother@dont-bother-replying-we-wont-read-it. com" isn't misleading at all, just non-helpful.
I've had times when that was the kind of work I was doing, and you're right on about it - and when I read the first paragraph or two I was thinking about Peopleware:-) At Bell Labs in the 70s-80s, we normally had two-person offices for most of the workers (1-person for supervisors, 3-person when we were short on space, or 4-5 person when we stole conference rooms for office space, but mostly 2-person.) It was usually quiet enough to focus, but social enough to communicate.
Different tasks have different needs, and different jobs have different mixtures of tasks. There are some tasks where you need peace and quiet for hours of uninterrupted concentration to be really productive - you keep the ideas you need in your head and get into a flow, and if that's interrupted a lot you lose productivity rapidly and spend all your time in slow-start, and you need all your books right at hand on your bookshelf. There are other tasks where your biggest needs are communication with your customer, so as long as you've got a cellphone and laptop it doesn't matter if you've got a desk. There are tasks where you need to be interacting a lot with the other members of your team, so cubicles or open lab-table space work fine. And there are tasks where you need hardware to mess with.
I started my career at Bell Labs in an environment with two-person offices, which was usually excellent for most work. It's quiet enough that you can think, but you're not isolated (so if you're the type who needs other people there to stay motivated, you've got that, and if you want to have conversations about what's going on you can.) If you needed to ignore everybody and focus, you went to the library.
The last 10 years, in different parts of AT&T, I've usually been in cubicles plus home office - the last six months I've mostly worked from home, because I'm on the phone a lot, email constantly, supporting people who are out of town, and I go to the office once or twice a month. But for the first year or so doing my current job, it was really helpful being in the office to absorb the relationships and corporate culture and politics and current events. That's one thing I miss when I'm working from home, but the office is usually relatively empty, and we're not hiring new people very often so the fact that I'm not there to help socialize them doesn't get lost. Part of the reason our San Francisco office is relatively empty is that it's expensive real estate, so the Corporate Real Estate Goons decided everybody should have small shared cubicles, and most of the people were sales people who can spend time at their customers' offices or work from home, so they got the hint and stopped showing up.
I was also going to post a suggestion about Apple putting dual jacks on the phone. But until they do, a Y-cable is a cheap and simple substitute.
BT Headsets Aren't Musical.
on
iPod-Jacked
·
· Score: 1
Bluetooth headsets are mostly designed for cell phones, not music - not only mono as opposed to stereo, but they're a pretty low frequency band (I think 3kHz or something.) A friend of mine at Apple says that they're too low-res to run Apple speech recognition on, unfortunately, because otherwise that BT-enabled 802.11-enabled Mac in your backpack would be a hands-free voice PDA plus VOIP-phone.
Disposable Income != Parking Convenience
on
iPod-Jacked
·
· Score: 1
Just because you've got money doesn't mean parking is convenient at home or near the office. (And if this was San Francisco, and you've got a van, parking is especially inconvenient:-)
Why? They don't have my real address. Because the main Yahoogroups account I've got is for reading yahoogroups on the web, and the email's pointed at my old excite.com address (which probably doesn't exist any more, but I certainly don't read it), while the yahoo mail account that's attached gets any groups announcement and yahoo-announcements and a small amount of real spam, and I clear it out occasionally but don't read it. I've got another yahoo account I use to send anonymous coward mail, which I only read when I'm expecting a response.
The report said the big losses were in electronics manufacturing and telecommunications. Telecom's no surprise - we were a big part of the crash, after being even more radically overoptimistic and overbuilt and overspent than the web+advertising+software game, but manufacturing is more interesting. It sounds like some of this is a statistical problem - the category sounds like it includes electronics manufacturing companies, so losing developers gets lumped together with losing physical assembly work because customers aren't buying products, or moving the labor offshore or replacing people with robots. On the other hand, some of the commentary I've read on this sounds like the US is losing some of the final assembly work as well as the low-level component board work.
Crossover is endemic in this game. How many times have you seen stories bounce back and forth between Slashdot, Dave Farber's Intereting People list, Declan McCullagh's Politech list, EFF bulletins, NYTimes articles, FoRK, the blogiverse, Cypherpunks back when it was more active, etc.? The important thing is to keep track of the source enough to notice whether a story is new, or just a rerun, and whether two stories that seem to reinforce each other are just different retellings of the same source.
This was earlier in the Crash, so after she'd been laid off with notice from her previous company, she'd accepted the best-looking of the jobs she could find. A week later she got laid off in the morning, and was working at the second-best-looking job in the afternoon.
Multicast is available in IPv4, but not too many ISPs support it natively, and I'm not aware of any of the major ISPs doing multicast peering with each other except to the extent that they connect to the mbone. There has been lots of work on reliable multicast for file distribution, but that's a lot more useful in a controlled environment, e.g. a many-site company distributing software updates or retail price lists over the Internet instead of satellite, than they are in random crowds where people join and leave. Unreliable multicast applications are a lot easier to coordinate over random crowds, e.g. video conferences, where if you miss a few packets you're not going to go back and recover them.
There are 43.5 million Koreans, so that's rather more money per household, i.e. per connection, and maybe not all of them would want to pay that much if they had a choice. Also, a high fraction of the Korean population live in large apartment buildings, where there's a huge economy of scale possible (which is why so many people there have 10 Mbps or other high-speed service.) There's also a lot of rural and mountainous country, where that kind of service may not be realistic.
OK, maybe it is spread over five years, but that's still the kind of pork barrel you get when something's being proposed more for political image than actual economics.
You're misunderstanding IPv6, but that's not relevant here because this project isn't about IPv6. IPv6 does have enough address space that it's possible to do some consolidated routing tables, especially for lower-level nodes, but if the network topology isn't tree-structured (and it almost never is), then a tree-structured search is unlikely to get you an optimal route.
This project seems to be about two things
A distributed bunch of fat servers around the world (well, mostly North America and Europe) that academically correct data publishers can upload their content to, with N copies of everything they're publishing spread around different servers, on the Internet 2 fat underloaded academic backbone.
A protocol for striped downloads from multiple servers, so if a given server is overloaded and slow it doesn't become the bottleneck, plus windowing download features that can handle a bigger speed*delay cross product than TCP's rather limited single-stream windowing can.
Anybody who wants to publish something reliably just needs to put up a seed and keep it up. If they want to publish something _more_ reliably, they can put up two seeds in different places to decrease the chance of failure, plus if they're distributing something with worldwide popularity, it helps to put seeds on several continents. if a given file becomes less popular, on a server which also has material that's more popular, the possible download rates obviously go way down, but there's no reason for something you're publishing to go unfinishable if you don't want it to.
People who only half-assedly care about publishing their material don't need to do this, of course, and your experience with stuff getting aborted in the middle is probably for popular material like anime or perhaps concert recordings where the people who publish it take it down when it's no longer "current". There's no reason that this has to happen for your own garage band's main CDs, even though you might not keep every show on line permanently.
On the other hand, an obvious feature for BitTorrent to add, if it doesn't have this already, is a "De-Publish" option which will let people who have part of the file get the rest of it but won't serve new download sessions to people who haven't already started. That'd be enough to give a day's warning when taking something down. You could either implement it mandatorily in the server or permissively in the client (e.g. make the server enforce it, which is more work, or let the client tell the user that the flag is set for 3.27 hours from now [Continue?][GiveUp?].
Doing lots of Pentium-specific tweaks in your code may be zero or negative help when somebody wants to run it on a Macintosh....
More importantly, tweaking code for heavy optimization is not usually a good job for humans. It's fine if you've got a piece of hardware that you have to tweak the last few percent performance out of for an application that will run for many CPU-years, but you're almost always much better off if you
Pick good algorithms
Leave bit-bashing loop-twiddly optimization tricks to the compiler, and
Put your human creativity skills to work on the next revision of your code, or on your next project if this one is done.
There are kinds of flaws that chroot jails can defend against, and it's certainly worth building separate environments for different applications to limit the effects of attacks, but that won't help you for this kind of attack (e.g. a similar abuse of sendmail.) The problem is that the miscreant can talk to the email server and ask it to forward mail, all of which the email server is doing under its own privileges - chroot can prevent the email system from being used to attack DNS, but not this. Similarly, the SQL Slammer worm was a very clever hack that took the SQL server and tricked it into sending out packets as itself - a chroot equivalent for MS wouldn't have stopped that either.
OK, I didn't quite leave it parked in Manhattan with a big "Steal Me" sign on it, but it wouldn't have broken my heart if my 150,000-mile rusty Ford had gotten stolen back in the mid-80s:-) Actually, somebody did break into it in the train station in New Jersey and broke the dashboard while unsuccessfully trying to steal the Ford OEM boring radio, and the $180 I got from the insurance was more than I eventually got from selling the car...
Slashdot Mirror
Sure, there are applications for which total
portability and a USB key are fine. But often I'd like to be able to work with files on a standalone machine, access downloaded material, etc., even though the underlying machine is configured for Windows. It'd be especially nice to have features like recognizing a
Another environment for which this is useful is little server machines hanging off an Internet connection where you're not monitoring security all the time. The thing can sit around serving web pages, running intrusion detection, acting as a honeypot or teergrube for annoying spammers, etc., and if it gets attacked and rooted, you just reboot. Perhaps a more security-oriented distro would be better for this, since this sounds more desktop-oriented, but it'd be nice to have one consistent environment.
Once you pull the pin, Mr. Grenade is no longer your friend... That _was_ the patriotic pin you were talking about, wasn't it?
But apparently the Bush Administration is looking for new recruits for the Draft Board. They haven't said when the Homeland Security Patriotic Anti-Terrorist Force starts. Yet.
But just as you can write undecipherable undocumented bad spaghetti code in almost any computer or non-computer language, you can also write uselessly non-standardizable XML that nobody can communicate with independently, even if you _aren't_ deliberately trying to obfuscate it. Microsoft Office's latest version uses XML for data interchange between components, and I've heard people assert that it's not possible for a programmer who isn't buying a bunch of APIs (or otherwise getting Office to do tasks) to communicate with it dependably. I haven't verified that myself, but I've seen the HTML code produced by previous versions, so I'm inclined to believe it.
Sure, XML isn't inherently that deep - but neither are the tab-separated ASCII files which Unix tools used to do all kinds of really powerful things. Similarly, LISP property lists aren't that complex. XML's a bit more flexible, and carries enough decoration with it that people are willing to use it for building interfaces that they might not build using ASCII or XDR. And anything that lets the EDI people replace their stuff with simpler, more open technology is good too..
On the other hand, during the 80s disco years, the profit on importing coke was in the $1-10M per Cessna-load range, so you could afford to lose half your shipments, and there were plenty of pilots willing to risk a few years in jail for a share of the profits if they win. The profit margin's probably not still 90%, but that's mostly because it's so easy for the importers; even at 50% you can lose almost half your shipments, so as long as you've got enough capital to risk losing a couple of containerloads before one gets through, your real problem becomes making reliable connections with distributors and not getting ripped off on delivery.
The ban on "misleading" headers is unclearly worded, but appears to ban sending email without headers that are actively helpful to "law enforcement". It's not clear whether this bans anonymous remailers in the US, because some of the mail they remove headers from _could_ be from spammers, even though spam is an abuse of what they're intended for. It does look like the bureaucrats (mainly FTC) who implement the policies will have enough definition-making ability that they could probably treat it that way.
In general, the bill fails to differentiate between forging a From or contact email address that's a real email address belonging to someone else (whether joe-job or random), forging a "nobody"-type address at a real domain belonging to someone else, forging a fake address at a real domain belonging to someone else, using a fake address at a real domain that the spammer's allowed to use, using an address at a fake domain, or other things. Forging the address of a real person is highly annoying to them, and might already legally count as forgery. Forging yet another bogus hotmail address is annoying to hotmail, but less annoying than a real address; using "dont-bother@dont-bother-replying-we-wont-read-it. com" isn't misleading at all, just non-helpful.
I've had times when that was the kind of work I was doing, and you're right on about it - and when I read the first paragraph or two I was thinking about Peopleware :-) At Bell Labs in the 70s-80s, we normally had two-person offices for most of the workers (1-person for supervisors, 3-person when we were short on space, or 4-5 person when we stole conference rooms for office space, but mostly 2-person.) It was usually quiet enough to focus, but social enough to communicate.
There are some tasks where you need peace and quiet for hours of uninterrupted concentration to be really productive - you keep the ideas you need in your head and get into a flow, and if that's interrupted a lot you lose productivity rapidly and spend all your time in slow-start, and you need all your books right at hand on your bookshelf. There are other tasks where your biggest needs are communication with your customer, so as long as you've got a cellphone and laptop it doesn't matter if you've got a desk. There are tasks where you need to be interacting a lot with the other members of your team, so cubicles or open lab-table space work fine. And there are tasks where you need hardware to mess with.
I started my career at Bell Labs in an environment with two-person offices, which was usually excellent for most work. It's quiet enough that you can think, but you're not isolated (so if you're the type who needs other people there to stay motivated, you've got that, and if you want to have conversations about what's going on you can.) If you needed to ignore everybody and focus, you went to the library.
The last 10 years, in different parts of AT&T, I've usually been in cubicles plus home office - the last six months I've mostly worked from home, because I'm on the phone a lot, email constantly, supporting people who are out of town, and I go to the office once or twice a month. But for the first year or so doing my current job, it was really helpful being in the office to absorb the relationships and corporate culture and politics and current events. That's one thing I miss when I'm working from home, but the office is usually relatively empty, and we're not hiring new people very often so the fact that I'm not there to help socialize them doesn't get lost. Part of the reason our San Francisco office is relatively empty is that it's expensive real estate, so the Corporate Real Estate Goons decided everybody should have small shared cubicles, and most of the people were sales people who can spend time at their customers' offices or work from home, so they got the hint and stopped showing up.
I was also going to post a suggestion about Apple putting dual jacks on the phone. But until they do, a Y-cable is a cheap and simple substitute.
Bluetooth headsets are mostly designed for cell phones, not music - not only mono as opposed to stereo, but they're a pretty low frequency band (I think 3kHz or something.) A friend of mine at Apple says that they're too low-res to run Apple speech recognition on, unfortunately, because otherwise that BT-enabled 802.11-enabled Mac in your backpack would be a hands-free voice PDA plus VOIP-phone.
Just because you've got money doesn't mean parking is convenient at home or near the office. (And if this was San Francisco, and you've got a van, parking is especially inconvenient :-)
Why? They don't have my real address. Because the main Yahoogroups account I've got is for reading yahoogroups on the web, and the email's pointed at my old excite.com address (which probably doesn't exist any more, but I certainly don't read it), while the yahoo mail account that's attached gets any groups announcement and yahoo-announcements and a small amount of real spam, and I clear it out occasionally but don't read it. I've got another yahoo account I use to send anonymous coward mail, which I only read when I'm expecting a response.
The report said the big losses were in electronics manufacturing and telecommunications. Telecom's no surprise - we were a big part of the crash, after being even more radically overoptimistic and overbuilt and overspent than the web+advertising+software game, but manufacturing is more interesting. It sounds like some of this is a statistical problem - the category sounds like it includes electronics manufacturing companies, so losing developers gets lumped together with losing physical assembly work because customers aren't buying products, or moving the labor offshore or replacing people with robots. On the other hand, some of the commentary I've read on this sounds like the US is losing some of the final assembly work as well as the low-level component board work.
Crossover is endemic in this game. How many times have you seen stories bounce back and forth between Slashdot, Dave Farber's Intereting People list, Declan McCullagh's Politech list, EFF bulletins, NYTimes articles, FoRK, the blogiverse, Cypherpunks back when it was more active, etc.? The important thing is to keep track of the source enough to notice whether a story is new, or just a rerun, and whether two stories that seem to reinforce each other are just different retellings of the same source.
This was earlier in the Crash, so after she'd been laid off with notice from her previous company, she'd accepted the best-looking of the jobs she could find. A week later she got laid off in the morning, and was working at the second-best-looking job in the afternoon.
Multicast is available in IPv4, but not too many ISPs support it natively, and I'm not aware of any of the major ISPs doing multicast peering with each other except to the extent that they connect to the mbone. There has been lots of work on reliable multicast for file distribution, but that's a lot more useful in a controlled environment, e.g. a many-site company distributing software updates or retail price lists over the Internet instead of satellite, than they are in random crowds where people join and leave. Unreliable multicast applications are a lot easier to coordinate over random crowds, e.g. video conferences, where if you miss a few packets you're not going to go back and recover them.
OK, maybe it is spread over five years, but that's still the kind of pork barrel you get when something's being proposed more for political image than actual economics.
This project seems to be about two things
People who only half-assedly care about publishing their material don't need to do this, of course, and your experience with stuff getting aborted in the middle is probably for popular material like anime or perhaps concert recordings where the people who publish it take it down when it's no longer "current". There's no reason that this has to happen for your own garage band's main CDs, even though you might not keep every show on line permanently.
On the other hand, an obvious feature for BitTorrent to add, if it doesn't have this already, is a "De-Publish" option which will let people who have part of the file get the rest of it but won't serve new download sessions to people who haven't already started. That'd be enough to give a day's warning when taking something down. You could either implement it mandatorily in the server or permissively in the client (e.g. make the server enforce it, which is more work, or let the client tell the user that the flag is set for 3.27 hours from now [Continue?][GiveUp?].
More importantly, tweaking code for heavy optimization is not usually a good job for humans. It's fine if you've got a piece of hardware that you have to tweak the last few percent performance out of for an application that will run for many CPU-years, but you're almost always much better off if you
Kilometers? In my day, when Americans went to the moon, we only had *miles*, and only had 238,000 of them....
Alternatively, go see Moonie and Broon.
There are kinds of flaws that chroot jails can defend against, and it's certainly worth building separate environments for different applications to limit the effects of attacks, but that won't help you for this kind of attack (e.g. a similar abuse of sendmail.) The problem is that the miscreant can talk to the email server and ask it to forward mail, all of which the email server is doing under its own privileges - chroot can prevent the email system from being used to attack DNS, but not this. Similarly, the SQL Slammer worm was a very clever hack that took the SQL server and tricked it into sending out packets as itself - a chroot equivalent for MS wouldn't have stopped that either.
OK, I didn't quite leave it parked in Manhattan with a big "Steal Me" sign on it, but it wouldn't have broken my heart if my 150,000-mile rusty Ford had gotten stolen back in the mid-80s :-) Actually, somebody did break into it in the train station in New Jersey and broke the dashboard while unsuccessfully trying to steal the Ford OEM boring radio, and the $180 I got from the insurance was more than I eventually got from selling the car...