While I don't disagree about Bernstein's personality or coding, um, "style", it should be noted that many of his "random" calls noted above are really intended to be "unique" not "random" despite the variable name. Time + pid is a reasonable way to ensure that you get a unique file name, at least on systems that don't recycle PID's more than once a second. If you actually generated a random file name it would probably be unique, but it's not guarateed to be so, which requires file locks, which is complicated on NFS for example.
Likewise the timeout handling is another place where "random" doesn't mean random so much as "unique amongst currently running qmail processes" to avoid having more than one concurrent process timeout at exactly the same time. getpid() is cheaper than rand() and doesn't eat from your entropy. It's poorly named, and without comments it's most certainly unclear, but it's not necessarily bad from a functional standpoint.
That's GSM for you, if you want the one-number-per-contact, no-data-other-than-phone address book that you can write to any GSM SIM I've ever had. I have waaaay too many contacts to deal with a system like that.
Oh, and Sun wouldn't have had this problem if they'd used pure Java code rather than relying on an existing library.
Yes. If only Sun would toss out all the C and re-implement their JVM in Java. How you'd launch the Java-based JVM is not clear, but once you got it going you'd never have to worry about buffer overflows again.
And your photographic copy would A) degrade over time and B) lose quality with each copy. IMHO that's not a very good archive. Moreover, in order to slow the inevitable decay that comes with time and reactive chemicals on paper/plastic/metal/whatever, you'd still need a climate-controlled facility. And you'd still need a team of operators to make the copy, and to make later copies as the earlier ones degrade. And more than anything else, you'd need someplace to store *another* 165 tons of photos, which is certainly larger than the space required to store a petabyte of data in a modern digital format.
I'm not really seeing how your photographic archive saves money. I'm not convinced it would produce better longevity either. You might get better longevity for a single copy than with digital data, but it's a whole lot cheaper to make digital copy #2 than to make photographic copy #2.
If you're worried about file formats you could simply leave a printed text detailing the data format. Then anyone with the ability to read the media would be able to recreate viewing software, even if none existed for then-modern computers.
If you're worried about being able to read the media then you're really worried about ongoing funding -- someone to continue preserving the archive in the future. That's a problem that exists regardless of the format of the archive; if someone decided they didn't want to keep paying for 3 floors of a building, or to continue making copies of the photographic archive, you'd still be in trouble.
Last time I tried to get a SIM the easiest course was to simply take the "free" phone and steal the SIM from it. At the time that didn't have any penalty in terms of extending my (already existing) contract, but they weren't even slightly interested in simply sending me a SIM. I've heard that T-Mobile will send you a SIM so long as you've already got a contract, but I've had no such luck with AT&T.
And good luck trying to bring your existing equipment to a any new provider without signing up for the same 2-year contract you'd get with your subsidized equipment -- you're paying them for the free phone, including the cancellation fee, even if you don't get one.
One other point is that while things like cameras and checking ID may not always deter or prevent a crime or an attack, it often greatly assists in the investigation after the fact.
This is my problem. They don't actually stop crime, they just make it cheaper to prosecute. And if cheaper prosecution is the only justification we need for reduced privacy, where do we draw the line? Improved prosecution does not significantly improve safety -- the UK has much better conviction rates with cameras, but crime rates are unaffected. Since society gains no security society should not be willing to trade much of anything (privacy, freedom, etc.) for this "benefit".
I personally want prosecution to be somewhat difficult and expensive; when it becomes cheap to simply ticket people using "irrefutable" automated evidence collection system there's a huge temptation to turn criminal laws sources for revenue. We've already seen it happen with automated traffic ticketing systems -- are you ready to be sent a use tax bill for items that were automatically determined to be in your car when you entered the state but not when you left, including fuel?
And that's one of the "legal" uses of the system -- I haven't even touched on the illegal abuses of power that a wide-scale automated evidence gathering system is sure to create. If you've ever meet a human you should be aware that people *will* abuse this sort of system. Prosecutorial and police malfeasance has and will continue to occur; even if abuse is rare, I still don't see what we're getting that's worth the additional risk.
IP-layer encryption doesn't do ports, period -- that's a transport-layer concept. And since IP-layer essentially implies host-level encryption, there's no need to multiplex, so it would be unecessary anyway.
It's also not an issue for end-host firewalls, since a transport-layer-aware firewall would get the decrpyted traffic with plain old TCP in it away; seeing the port is only a problem for in-line firewalls. And you could solve that problem by decrypting at the border and forwarding un-encrypted (or re-encrypted if you like) packets to the end-host. You'd have to ensure that your key negotiation allowed for this, but from a technical perspective it's not terribly complicated; it's a lot like the standard tunneling-concentrator VPN setup that many people use.
You can know what's encrypted and what's not because whatever host does the decryption either does or doesn't decrypt things, and it knows whether or not that decryption step took place. A simple setup like this: Internet->Firewall (blocks non-encrypted IP packets)->Server->Decrypt->Local Firewall (blocks bad ports)->Upper layers would be entirely possible with today's IPSEC technology and firewalls, and would require no modification of any upper-layer protocols.
Given a 8-foot stick, a string, and a knife, accurately measure 1 inch. Use a calculator if you like: spaceheater ~ 0$ factor 96 96: 2 2 2 2 2 3
Now, given a 1-meter stick a string, and a knife, accurately measure 1 centimeter. Use a calculator if you like: spaceheater ~ 0$ factor 100 100: 2 2 5 5
It's a lot easier to divide a string accurately by 2 (or 4 or 8) or even 3 than by 5.
The hard to replace battery is not necessarily "planned obsolescence" if Apple designed with the thought that most people would replace their phone before the battery was dead. The term "planned obsolescence" implies intent -- intent that would be absent if they didn't think the battery would be an issue for must users.
I'm an Apple apologist; it's quite possible that Apple *did* intend to make the battery a limiting factor. But the fact that it's hard to replace is not sufficient evidence to demonstrate that intent -- it's equally possible that they figured the phone would be out of most user's lives before the battery died, and therefore choose the esthetic and structurally superior sealed case over the ability to easily replace the battery, without any thought of increasing sales through obsolescence.
"Visual voicemail" really doesn't require that much vendor support, at least not from a technical perspective. Any modern voicemail server has an option to attach voicemail to an email message and send it off. If your phone supports the native recording format of the voicemail system that's all you need -- an email. But even if you need to do some translation it's pretty easy; email->procmail->lame->MMS gateway == visual voicemail. I use exactly that system to get "visual voicemail" on my phone.
Now, I'm sure that providers aren't interested in letting you actually use the voicemail you're buying, for all sorts of reasons that I'm sure make sense in their 1970's telcom minds, but from a technical perspective it's trivial. Almost any modern phone can accept messages that include some sort of audio file, and your provider has all the information and technology necessary to send such messages to your phone when new voicemail arrives.
I'm not bashing the iPhone, but I'm sick of people pretending that "visual voicemail" is something complicated or new.
The traffic can be identified with some accuracy, but you still can't read the content. And if you can't read the content you can't do acoustic fingerprinting on the media files, which is what they've been ordered to do.
Instead of re-writing every protocol to look like IPSEC, couldn't we add a layer to the network stack between the transport layer and the IP layer to encrypt the IP payload? Then we wouldn't have to re-write all our old apps, wouldn't need to implement encryption in every app, and wouldn't need to try to hide the port numbers. If only there were such an IP-layer SECurity service...
Did you try talking to the hospital -- many are willing to give *huge* discounts if you offer to pay in-full without an insurance claim. They'll often knock 20%-30% off without even haggling, and there's still from room for negotiation from there. Between the mandatory discounts hospitals offer to insurance companies and the number of bills that never get paid list prices for hospital services can be quite high, but at least in my experience there's a lot of room to recover some of that markup, because the hospital didn't really expect to get that much money in the first place.
He never suggested that his old phone was better, just that cellular reception was poor in is home. It's possible that the AT&T GSM signal is worse, but his post didn't tell us that. Maybe you should explore posts before insulting people.
He didn't say it would do the aliens an enormous amount of good, but that it would do *us* an enormous amount of good. Having a universal enemy would go a long way to give people some perspective on things like dogs, cats, blacks, gays, lesbians, arabs and the other unimportant things that we usually fight about.
Plus who says the aliens aren't here to steal our water/children/etc. anyway? Just because you aspire to be tolerant and social doesn't mean aliens would be. I think it's both arrogant and optimistic to assume that other lifeforms would be interested in our life in any way that would allow civil interaction, regardless of how humans react -- aliens might simply not care to interact with us, might consider us a threat, or might just be coming through with their constructor fleet and find our planet in the way.
Most routers -- particularly those on the public Internet -- route only at the IP level and not at the TCP level. As such they would not notice anything different about these packets.
Well now he's not -- I like the nub mouse too, particular if I'm switching between typing and pointing, as there's less hand movement required to switch.
Trackpads today are useable (though still not my preference), but when they first started replacing working pointing interfaces -- like the nub mouse -- with those worthless first and second gen trackpads I would have given anything for the nub mouse.
No, it's not, because "planned" requires intent. If they really thought that consumer interest would be the limiting factor in the life of the phone then failing to design it function longer than its expected lifetime does not show intent -- planning -- to limit that lifetime. I'm not saying that's how it happened, but you can't claim "planned obsolescence" without implying intent.
Your tires have a shorter lifetime than most other parts in your car and aren't user-replacable -- you have to go someplace with specialized equipment to rip the old tires off your rims and install new ones. Is that "planned obsolescence" or does it simply reflect that fact that both tires and cars can be superior (either in price or engineering or both) if we don't make them user-replacable?
Yes, just like speeding tickets keep people from speeding.
You can't deter someone from an action unless they think they'll be accountable to your deterrent -- if someone commits a crime expecting to get away with it the penalty is irrelevant.
For one thing, just because it runs that way doesn't mean it was re-designed to take advantage of that fact. Any current OS *could* run on a system composed entirely of fast, solid-state memory. But to take advantage of that you'd have to make changes to do things like in-place execution and a new execution-space memory management scheme.
Even if Windows Mobile was designed with that in mind, it doesn't support things like huge address spaces or swap (not that you'd use swap directly, but you still need a scheme for temporary allocation of memory used in the execution of programs) or a variety of other features common in modern operating systems that you'd really miss, and which would also have to be re-designed to take advantage of the high-speed persistent storage.
Your claim is a little like one that a 1980's calculator wristwatch has a real-time OS designed from the ground-up to use solid-state components both for execution and long-term storage (the memory key). It's technically true, but it's meaninless in a discussion about general-purpose computers.
You had a commercially available computer before 1984 that had a mouse and a bit-map display and folders and icons? Or you got one in 1984 that wasn't from Apple?
Slashdot Mirror
While I don't disagree about Bernstein's personality or coding, um, "style", it should be noted that many of his "random" calls noted above are really intended to be "unique" not "random" despite the variable name. Time + pid is a reasonable way to ensure that you get a unique file name, at least on systems that don't recycle PID's more than once a second. If you actually generated a random file name it would probably be unique, but it's not guarateed to be so, which requires file locks, which is complicated on NFS for example.
Likewise the timeout handling is another place where "random" doesn't mean random so much as "unique amongst currently running qmail processes" to avoid having more than one concurrent process timeout at exactly the same time. getpid() is cheaper than rand() and doesn't eat from your entropy. It's poorly named, and without comments it's most certainly unclear, but it's not necessarily bad from a functional standpoint.
That's GSM for you, if you want the one-number-per-contact, no-data-other-than-phone address book that you can write to any GSM SIM I've ever had. I have waaaay too many contacts to deal with a system like that.
Oh, and Sun wouldn't have had this problem if they'd used pure Java code rather than relying on an existing library.
Yes. If only Sun would toss out all the C and re-implement their JVM in Java. How you'd launch the Java-based JVM is not clear, but once you got it going you'd never have to worry about buffer overflows again.
And your photographic copy would A) degrade over time and B) lose quality with each copy. IMHO that's not a very good archive. Moreover, in order to slow the inevitable decay that comes with time and reactive chemicals on paper/plastic/metal/whatever, you'd still need a climate-controlled facility. And you'd still need a team of operators to make the copy, and to make later copies as the earlier ones degrade. And more than anything else, you'd need someplace to store *another* 165 tons of photos, which is certainly larger than the space required to store a petabyte of data in a modern digital format.
I'm not really seeing how your photographic archive saves money. I'm not convinced it would produce better longevity either. You might get better longevity for a single copy than with digital data, but it's a whole lot cheaper to make digital copy #2 than to make photographic copy #2.
If you're worried about file formats you could simply leave a printed text detailing the data format. Then anyone with the ability to read the media would be able to recreate viewing software, even if none existed for then-modern computers.
If you're worried about being able to read the media then you're really worried about ongoing funding -- someone to continue preserving the archive in the future. That's a problem that exists regardless of the format of the archive; if someone decided they didn't want to keep paying for 3 floors of a building, or to continue making copies of the photographic archive, you'd still be in trouble.
Last time I tried to get a SIM the easiest course was to simply take the "free" phone and steal the SIM from it. At the time that didn't have any penalty in terms of extending my (already existing) contract, but they weren't even slightly interested in simply sending me a SIM. I've heard that T-Mobile will send you a SIM so long as you've already got a contract, but I've had no such luck with AT&T.
And good luck trying to bring your existing equipment to a any new provider without signing up for the same 2-year contract you'd get with your subsidized equipment -- you're paying them for the free phone, including the cancellation fee, even if you don't get one.
One other point is that while things like cameras and checking ID may not always deter or prevent a crime or an attack, it often greatly assists in the investigation after the fact.
This is my problem. They don't actually stop crime, they just make it cheaper to prosecute. And if cheaper prosecution is the only justification we need for reduced privacy, where do we draw the line? Improved prosecution does not significantly improve safety -- the UK has much better conviction rates with cameras, but crime rates are unaffected. Since society gains no security society should not be willing to trade much of anything (privacy, freedom, etc.) for this "benefit".
I personally want prosecution to be somewhat difficult and expensive; when it becomes cheap to simply ticket people using "irrefutable" automated evidence collection system there's a huge temptation to turn criminal laws sources for revenue. We've already seen it happen with automated traffic ticketing systems -- are you ready to be sent a use tax bill for items that were automatically determined to be in your car when you entered the state but not when you left, including fuel?
And that's one of the "legal" uses of the system -- I haven't even touched on the illegal abuses of power that a wide-scale automated evidence gathering system is sure to create. If you've ever meet a human you should be aware that people *will* abuse this sort of system. Prosecutorial and police malfeasance has and will continue to occur; even if abuse is rare, I still don't see what we're getting that's worth the additional risk.
IP-layer encryption doesn't do ports, period -- that's a transport-layer concept. And since IP-layer essentially implies host-level encryption, there's no need to multiplex, so it would be unecessary anyway.
It's also not an issue for end-host firewalls, since a transport-layer-aware firewall would get the decrpyted traffic with plain old TCP in it away; seeing the port is only a problem for in-line firewalls. And you could solve that problem by decrypting at the border and forwarding un-encrypted (or re-encrypted if you like) packets to the end-host. You'd have to ensure that your key negotiation allowed for this, but from a technical perspective it's not terribly complicated; it's a lot like the standard tunneling-concentrator VPN setup that many people use.
You can know what's encrypted and what's not because whatever host does the decryption either does or doesn't decrypt things, and it knows whether or not that decryption step took place. A simple setup like this:
Internet->Firewall (blocks non-encrypted IP packets)->Server->Decrypt->Local Firewall (blocks bad ports)->Upper layers
would be entirely possible with today's IPSEC technology and firewalls, and would require no modification of any upper-layer protocols.
Given a 8-foot stick, a string, and a knife, accurately measure 1 inch. Use a calculator if you like:
spaceheater ~ 0$ factor 96
96: 2 2 2 2 2 3
Now, given a 1-meter stick a string, and a knife, accurately measure 1 centimeter. Use a calculator if you like:
spaceheater ~ 0$ factor 100
100: 2 2 5 5
It's a lot easier to divide a string accurately by 2 (or 4 or 8) or even 3 than by 5.
The hard to replace battery is not necessarily "planned obsolescence" if Apple designed with the thought that most people would replace their phone before the battery was dead. The term "planned obsolescence" implies intent -- intent that would be absent if they didn't think the battery would be an issue for must users.
I'm an Apple apologist; it's quite possible that Apple *did* intend to make the battery a limiting factor. But the fact that it's hard to replace is not sufficient evidence to demonstrate that intent -- it's equally possible that they figured the phone would be out of most user's lives before the battery died, and therefore choose the esthetic and structurally superior sealed case over the ability to easily replace the battery, without any thought of increasing sales through obsolescence.
"Visual voicemail" really doesn't require that much vendor support, at least not from a technical perspective. Any modern voicemail server has an option to attach voicemail to an email message and send it off. If your phone supports the native recording format of the voicemail system that's all you need -- an email. But even if you need to do some translation it's pretty easy; email->procmail->lame->MMS gateway == visual voicemail. I use exactly that system to get "visual voicemail" on my phone.
Now, I'm sure that providers aren't interested in letting you actually use the voicemail you're buying, for all sorts of reasons that I'm sure make sense in their 1970's telcom minds, but from a technical perspective it's trivial. Almost any modern phone can accept messages that include some sort of audio file, and your provider has all the information and technology necessary to send such messages to your phone when new voicemail arrives.
I'm not bashing the iPhone, but I'm sick of people pretending that "visual voicemail" is something complicated or new.
The traffic can be identified with some accuracy, but you still can't read the content. And if you can't read the content you can't do acoustic fingerprinting on the media files, which is what they've been ordered to do.
Instead of re-writing every protocol to look like IPSEC, couldn't we add a layer to the network stack between the transport layer and the IP layer to encrypt the IP payload? Then we wouldn't have to re-write all our old apps, wouldn't need to implement encryption in every app, and wouldn't need to try to hide the port numbers. If only there were such an IP-layer SECurity service...
It's not just you.
Did you try talking to the hospital -- many are willing to give *huge* discounts if you offer to pay in-full without an insurance claim. They'll often knock 20%-30% off without even haggling, and there's still from room for negotiation from there. Between the mandatory discounts hospitals offer to insurance companies and the number of bills that never get paid list prices for hospital services can be quite high, but at least in my experience there's a lot of room to recover some of that markup, because the hospital didn't really expect to get that much money in the first place.
unless acted upon by an outside force
An outside force you say. Would someone trying to steal its kinetic energy to generate energy possibly be such a force?
Just wondering.
He never suggested that his old phone was better, just that cellular reception was poor in is home. It's possible that the AT&T GSM signal is worse, but his post didn't tell us that. Maybe you should explore posts before insulting people.
He didn't say it would do the aliens an enormous amount of good, but that it would do *us* an enormous amount of good. Having a universal enemy would go a long way to give people some perspective on things like dogs, cats, blacks, gays, lesbians, arabs and the other unimportant things that we usually fight about.
Plus who says the aliens aren't here to steal our water/children/etc. anyway? Just because you aspire to be tolerant and social doesn't mean aliens would be. I think it's both arrogant and optimistic to assume that other lifeforms would be interested in our life in any way that would allow civil interaction, regardless of how humans react -- aliens might simply not care to interact with us, might consider us a threat, or might just be coming through with their constructor fleet and find our planet in the way.
Most routers -- particularly those on the public Internet -- route only at the IP level and not at the TCP level. As such they would not notice anything different about these packets.
Well now he's not -- I like the nub mouse too, particular if I'm switching between typing and pointing, as there's less hand movement required to switch.
Trackpads today are useable (though still not my preference), but when they first started replacing working pointing interfaces -- like the nub mouse -- with those worthless first and second gen trackpads I would have given anything for the nub mouse.
No, it's not, because "planned" requires intent. If they really thought that consumer interest would be the limiting factor in the life of the phone then failing to design it function longer than its expected lifetime does not show intent -- planning -- to limit that lifetime. I'm not saying that's how it happened, but you can't claim "planned obsolescence" without implying intent.
Your tires have a shorter lifetime than most other parts in your car and aren't user-replacable -- you have to go someplace with specialized equipment to rip the old tires off your rims and install new ones. Is that "planned obsolescence" or does it simply reflect that fact that both tires and cars can be superior (either in price or engineering or both) if we don't make them user-replacable?
If there were an approved treatment for males that might make more sense.
Yes, just like speeding tickets keep people from speeding.
You can't deter someone from an action unless they think they'll be accountable to your deterrent -- if someone commits a crime expecting to get away with it the penalty is irrelevant.
Why doesn't that apply to individuals as well? Why are corporations different in the structure?
For one thing, just because it runs that way doesn't mean it was re-designed to take advantage of that fact. Any current OS *could* run on a system composed entirely of fast, solid-state memory. But to take advantage of that you'd have to make changes to do things like in-place execution and a new execution-space memory management scheme.
Even if Windows Mobile was designed with that in mind, it doesn't support things like huge address spaces or swap (not that you'd use swap directly, but you still need a scheme for temporary allocation of memory used in the execution of programs) or a variety of other features common in modern operating systems that you'd really miss, and which would also have to be re-designed to take advantage of the high-speed persistent storage.
Your claim is a little like one that a 1980's calculator wristwatch has a real-time OS designed from the ground-up to use solid-state components both for execution and long-term storage (the memory key). It's technically true, but it's meaninless in a discussion about general-purpose computers.
You had a commercially available computer before 1984 that had a mouse and a bit-map display and folders and icons? Or you got one in 1984 that wasn't from Apple?