Do you know what some of us more seasoned veterans in the industry call a passionate criticizer?
A volunteer [to fix the code].
Once where I worked, there was this horrible piece of user interface code (code that handles keystrokes always tends to be). And there was this programmer who was constantly flaming about how bad it was. Then there was the meeting with management about adding Kanji support to the operating system. He wasn't at the meeting, but everyone else familiar with the code was and got assigned to other tasks. The meeting finally got down to this one ugly piece of code and nobody left to enhance it. And somebody chimes in, "Mr X, was just talking about it the other day..." Note that "complaining" got translated to the less specific "talking". Mr X did a great job on that code, BTW. Nobody has ever complained about it since...
I believe that doing something like that would violate the Computer Fraud and Abuse Act. Even if the ISP does it. An ISP could shut off the connection, but they don't have the authority to hack into anybody's computer.
I would set up a honeypot instead of just bouncing the scans -- and find some way to claim damages from whatever they do after the scans.
Or, how about sending a registered letter to their legal department saying that your fee for processing port scans is $10 per packet. If they don't stop, then bill them. If they don't pay, then you have suffered a business loss and you can claim damages which might make the scans a felony under the Computer Fraud and Abuse Act.
In addition to things already mentioned (nyquist, speaker/mic response, mics not turned on) the really big reason this won't work is because almost all commercial audio transmission is highly compressed in the frequency domain. Inaudible frequencies got unceremoniously tossed to reduce bandwidth.
PIM stands for Protocol Independent Multicast. That means the protocol is the same thing whether routing IPv4 or IPv6. Those multicast routing protocols may well have problems, but the problems will be the same for IPv4 and IPv6. Not to say the routers and switches won't have bugs: I would definitely recommend updating the firmware in any routers and switches you run IPv6 on (or if you are using PIM with IPv4).
In a classic router, every packet hits the CPU, not just multi-casts. It is certainly possible that a high performance router could optimize IPv4 forwarding, but not Iv6 forwarding. That has nothing to do with the protocol being "experimental". FYI, most of the protocols defined by the IETF are "experimental".
Most decent NICs these days have large tables to sort out the multi-cast addresses before they interrupt the host. They operate on the MAC address mapping for the multicasts; so, they are essentially the same whether IPv4 or IPv6. I know FreeBSD hosts do properly fill in the hardware multicast address tables in the NICs the same way whether using IPv4 or IPv6. I would assume Linux does, too. In other words, if your multicast traffic is hitting the host CPU for every packet, you may be running crappy NICs.
I believe the 82576 supports 23 multicast addresses before overflowing (and going into promiscuous mode). The i350 supports 31. The 82599 and x540 support 127. Those are Intel parts, but I suspect other vendors are doing the same.
Also, most switches made in the last 10 years support MLD snooping and won't forward IPv6 multi-cast packets to every port. You might have to turn on the MLD snooping... One problem that could come up is exceeding the size of multicast forwarding tables in old switches (causing them to either forwarding multicasts to every port or to use imperfect hashing schemes).
Correct, but possibly confusing. IPv6 systems will NOT fall back to sending broadcast packets, but switches that don't support MLD will transmit multicast packets for IPv6 to all ports on the switch. Where switches are concerned, MLD (and IGMP for that matter) are optional optimizations for specific Ethertypes. Historically, IGMP was used by hosts to communicate with routers assuming that the subnet was a party line (because that's what early Ethernet was). When we transitioned from coax and hubs to switches, the switch manufacturers figured out how to use IGMP to optimize where multicast packets were sent -- this is called IGMP snooping. IGMP is used to optimize traffic for the IP Ethertype (0x800). Now, MLD does the same thing for the IPv6 Ethertype. All other Ethertypes are shipped to every port. Of course, whether the switch actually implemented MLD snooping is an issue, but that's not the protocol.
FYI, MLD is pretty much IGMP with larger addresses. There are other things in IPv6 that are really new, this isn't.
It might actually be a good idea for Google to comply (at least in some cases), because the search results would be more relevant in many (most) cases. Taking the one of the examples given by others, most people really aren't going to care if somebody declared bankruptcy 40 years ago; so, showing that in a search result would just be adding noise (which Google search results already have too much of). In other words, as long as they have to filter the right to be forgotten requests, they might as well use that labor to improve search results everywhere.
In making this a worldwide feature, they could add an option to the advanced search that would enable returning "right to be forgotten" results but make the searcher agree to a legal notice swearing they were not in the EU before the option could be enabled. At that point the searcher would be the one violating EU law the data protection law and also probably committing fraud.
I don't think this is a joke at all. You were the one trotting out the good and evil cartoon lines and making it into a joke.
I think that re-instating the Fairness Doctrine would take away the unfair advantage money brings to the table. Or to put it another way, a debate isn't fair when one person is talking through a 10KW sound system and the other person has nothing.
Frankly, I want that ultra-conservative (or ultra-liberal) to not spend money on an ad. I want them to demonstrate support for their cause by showing how many people actually agree with them.
I see your point (didn't for a second -- the humor would have been clearer if you quoted "good" and "evil").
But a lone ranger type who is promoting "good" won't worry about "evil" getting equal time, because "good always triumphs over evil in a fair fight", right kemo sabe?
To fix that, he would need to either amend the constitution, or replace a few Supreme Court justices. That is not something he can "do quickly and then resign".
Actually, there is something that can be done quickly. It doesn't even take an Act of Congress: Re-instate a variant the Fairness Doctrine. The variant would apply to all political advertising that wasn't authorized and paid for by an FEC candidate committee (or state equivalent). To summarize, The new rule would require the purchaser of advertising to pay for both their own add and a rebuttal add. It might also be nice to require the media displaying the adds to fact check both the original add and the response (and be able to bill the purchaser something for doing that, too).
The Fairness Doctrine has already passed Supreme Court review. It's also likely to continue to pass Supreme Court review, because it creates more free speech as well as increasing the quality of the speech (by providing fact checking).
So, why haven't we already done that? I have submitted the idea through channels and nothing has happened. Sorry for being cynical here, but could it be because the politicians who claim to want to fix Citizen's United have found it to be a fund-raising cash cow and don't really want to fix it?
As others have pointed out, the formal verification does make the software provably reliable, but does nothing to protect against hardware issues. Just as a datapoint, the Stratus VOS operating system has been checksumming at the block driver level since the OS was written in 1980. It has detected failures on every generation of hardware it has been used with since. Some of the failures we have seen: Undetected transient media errors (the error correction/checking isn't perfect); Flaky I/O busses; bugs in RAID firmware.
These real world failures will swamp the potential failures caused by the software not being verified.
From what I've observed (in public sources) the methodology is something like googling for words like "terrorist" and "Islam" and putting every name that comes up onto the list. It seems more likely that the list is way under 1% accurate.
This is actually something you configure on your own computer. The FreeBSD command for doing it is ip6addrctl. I'm not sure what it's called on other systems. Also, note that the FreeBSD man pages are vague. You will also need to rfc-3484 to actually be able to use it. The man pages say it affects source and destination address selection, but they don't say how. It has two effects:
1. The command controls the sort used for DNS results. The default is to sort IPv6 addresses first. Properly converted applications actually get a list of IP addresses back from DNS and they are supposed to try them in order until the get one that works.
2. The command also controls which source address is paired with a destination address. This is less important, but something one might occasionally want to do.
For most ISPs in the US, it probably makes more sense to sort IPv4 first. At least for the next few years.
Somebody has to log in to a rather large number of routers and switches to enable IPv6 and possibly do a firmware upgrade, first. In case you haven't done it, yet, upgrading the firmware on a router/switch is the computer maintenance equivalent of going to the dentist. Especially when you save your configuration and the new firmware doesn't recognize the config file...
BTW, my cell phone has an IPv6 address, too.
Not sure how true this is, but I have heard that some of the large phone companies have so much internal switching equipment that they have run out of internal IPv4 addresses for the switching equipment (not to mention end user devices).
Slashdot Mirror
Do you know what some of us more seasoned veterans in the industry call a passionate criticizer?
A volunteer [to fix the code].
Once where I worked, there was this horrible piece of user interface code (code that handles keystrokes always tends to be). And there was this programmer who was constantly flaming about how bad it was. Then there was the meeting with management about adding Kanji support to the operating system. He wasn't at the meeting, but everyone else familiar with the code was and got assigned to other tasks. The meeting finally got down to this one ugly piece of code and nobody left to enhance it. And somebody chimes in, "Mr X, was just talking about it the other day..." Note that "complaining" got translated to the less specific "talking". Mr X did a great job on that code, BTW. Nobody has ever complained about it since...
I believe that doing something like that would violate the Computer Fraud and Abuse Act. Even if the ISP does it. An ISP could shut off the connection, but they don't have the authority to hack into anybody's computer.
Mine too. SSD upgrade is really nice...
In order for this to be implemented in the real world, there will have to be a standard.
The standard will be subject to creative interpretation.
Vendors will compete on how good their creative interpretation is at getting you through the intersection faster.
In other words, if this happens, it will turn into a high tech version of the game kids call "chicken".
The details of his IRS audits (they don't audit you more than once unless that find a lot the first time).
How many illegal immigrants he has hired.
The salary distribution for employees at his companies (i.e., the kind of jobs he knows how to create).
The manufacturing country of origin for the various products he has sold over the years (OK, that's probably not secret, but might be hard to find).
He should ask the IRS and see if they are off 8 track mag tape, yet. [This is no a a joke, they last I heard, they were trying to get off mag tape.]
One can't really understand much about economics without understanding logarithms.
I would set up a honeypot instead of just bouncing the scans -- and find some way to claim damages from whatever they do after the scans.
Or, how about sending a registered letter to their legal department saying that your fee for processing port scans is $10 per packet. If they don't stop, then bill them. If they don't pay, then you have suffered a business loss and you can claim damages which might make the scans a felony under the Computer Fraud and Abuse Act.
And cell phones still have crappy voice quality compared to land lines.
I signed the petition with the following comment:
I think any symbol adopted as a Unicode standard will dilute any trademark held on the symbol to the point where the trademark is no longer valid.
Feel free to pile on...
My brother named my nephew William. :-))
In addition to things already mentioned (nyquist, speaker/mic response, mics not turned on) the really big reason this won't work is because almost all commercial audio transmission is highly compressed in the frequency domain. Inaudible frequencies got unceremoniously tossed to reduce bandwidth.
A fire hose spraying diluted Coke syrup ought to do the trick...
Or paintball guns...
Good practice for the guards, too.
PIM stands for Protocol Independent Multicast. That means the protocol is the same thing whether routing IPv4 or IPv6. Those multicast routing protocols may well have problems, but the problems will be the same for IPv4 and IPv6. Not to say the routers and switches won't have bugs: I would definitely recommend updating the firmware in any routers and switches you run IPv6 on (or if you are using PIM with IPv4).
In a classic router, every packet hits the CPU, not just multi-casts. It is certainly possible that a high performance router could optimize IPv4 forwarding, but not Iv6 forwarding. That has nothing to do with the protocol being "experimental". FYI, most of the protocols defined by the IETF are "experimental".
Most decent NICs these days have large tables to sort out the multi-cast addresses before they interrupt the host. They operate on the MAC address mapping for the multicasts; so, they are essentially the same whether IPv4 or IPv6. I know FreeBSD hosts do properly fill in the hardware multicast address tables in the NICs the same way whether using IPv4 or IPv6. I would assume Linux does, too. In other words, if your multicast traffic is hitting the host CPU for every packet, you may be running crappy NICs.
I believe the 82576 supports 23 multicast addresses before overflowing (and going into promiscuous mode). The i350 supports 31. The 82599 and x540 support 127. Those are Intel parts, but I suspect other vendors are doing the same.
Also, most switches made in the last 10 years support MLD snooping and won't forward IPv6 multi-cast packets to every port. You might have to turn on the MLD snooping... One problem that could come up is exceeding the size of multicast forwarding tables in old switches (causing them to either forwarding multicasts to every port or to use imperfect hashing schemes).
Correct, but possibly confusing. IPv6 systems will NOT fall back to sending broadcast packets, but switches that don't support MLD will transmit multicast packets for IPv6 to all ports on the switch. Where switches are concerned, MLD (and IGMP for that matter) are optional optimizations for specific Ethertypes. Historically, IGMP was used by hosts to communicate with routers assuming that the subnet was a party line (because that's what early Ethernet was). When we transitioned from coax and hubs to switches, the switch manufacturers figured out how to use IGMP to optimize where multicast packets were sent -- this is called IGMP snooping. IGMP is used to optimize traffic for the IP Ethertype (0x800). Now, MLD does the same thing for the IPv6 Ethertype. All other Ethertypes are shipped to every port. Of course, whether the switch actually implemented MLD snooping is an issue, but that's not the protocol.
FYI, MLD is pretty much IGMP with larger addresses. There are other things in IPv6 that are really new, this isn't.
It might actually be a good idea for Google to comply (at least in some cases), because the search results would be more relevant in many (most) cases. Taking the one of the examples given by others, most people really aren't going to care if somebody declared bankruptcy 40 years ago; so, showing that in a search result would just be adding noise (which Google search results already have too much of). In other words, as long as they have to filter the right to be forgotten requests, they might as well use that labor to improve search results everywhere.
In making this a worldwide feature, they could add an option to the advanced search that would enable returning "right to be forgotten" results but make the searcher agree to a legal notice swearing they were not in the EU before the option could be enabled. At that point the searcher would be the one violating EU law the data protection law and also probably committing fraud.
I don't think this is a joke at all. You were the one trotting out the good and evil cartoon lines and making it into a joke.
I think that re-instating the Fairness Doctrine would take away the unfair advantage money brings to the table. Or to put it another way, a debate isn't fair when one person is talking through a 10KW sound system and the other person has nothing.
Frankly, I want that ultra-conservative (or ultra-liberal) to not spend money on an ad. I want them to demonstrate support for their cause by showing how many people actually agree with them.
I see your point (didn't for a second -- the humor would have been clearer if you quoted "good" and "evil").
But a lone ranger type who is promoting "good" won't worry about "evil" getting equal time, because "good always triumphs over evil in a fair fight", right kemo sabe?
To fix that, he would need to either amend the constitution, or replace a few Supreme Court justices. That is not something he can "do quickly and then resign".
Actually, there is something that can be done quickly. It doesn't even take an Act of Congress: Re-instate a variant the Fairness Doctrine. The variant would apply to all political advertising that wasn't authorized and paid for by an FEC candidate committee (or state equivalent). To summarize, The new rule would require the purchaser of advertising to pay for both their own add and a rebuttal add. It might also be nice to require the media displaying the adds to fact check both the original add and the response (and be able to bill the purchaser something for doing that, too).
The Fairness Doctrine has already passed Supreme Court review. It's also likely to continue to pass Supreme Court review, because it creates more free speech as well as increasing the quality of the speech (by providing fact checking).
So, why haven't we already done that? I have submitted the idea through channels and nothing has happened. Sorry for being cynical here, but could it be because the politicians who claim to want to fix Citizen's United have found it to be a fund-raising cash cow and don't really want to fix it?
Don't forget Dark Star...
As others have pointed out, the formal verification does make the software provably reliable, but does nothing to protect against hardware issues. Just as a datapoint, the Stratus VOS operating system has been checksumming at the block driver level since the OS was written in 1980. It has detected failures on every generation of hardware it has been used with since. Some of the failures we have seen: Undetected transient media errors (the error correction/checking isn't perfect); Flaky I/O busses; bugs in RAID firmware.
These real world failures will swamp the potential failures caused by the software not being verified.
From what I've observed (in public sources) the methodology is something like googling for words like "terrorist" and "Islam" and putting every name that comes up onto the list. It seems more likely that the list is way under 1% accurate.
This is actually something you configure on your own computer. The FreeBSD command for doing it is ip6addrctl. I'm not sure what it's called on other systems. Also, note that the FreeBSD man pages are vague. You will also need to rfc-3484 to actually be able to use it. The man pages say it affects source and destination address selection, but they don't say how. It has two effects:
1. The command controls the sort used for DNS results. The default is to sort IPv6 addresses first. Properly converted applications actually get a list of IP addresses back from DNS and they are supposed to try them in order until the get one that works.
2. The command also controls which source address is paired with a destination address. This is less important, but something one might occasionally want to do.
For most ISPs in the US, it probably makes more sense to sort IPv4 first. At least for the next few years.
Somebody has to log in to a rather large number of routers and switches to enable IPv6 and possibly do a firmware upgrade, first. In case you haven't done it, yet, upgrading the firmware on a router/switch is the computer maintenance equivalent of going to the dentist. Especially when you save your configuration and the new firmware doesn't recognize the config file...
BTW, my cell phone has an IPv6 address, too.
Not sure how true this is, but I have heard that some of the large phone companies have so much internal switching equipment that they have run out of internal IPv4 addresses for the switching equipment (not to mention end user devices).
That audio clip is also taken out of context..