As far as I know that requires a DPGS like system on the track with extra real time feedback to the car. So they are cheating if you consider the real world.
I've been in a car that could drive its self on one very well surveyed road. If it got confused it would beep and assume the human was in control within a second. The internal guidance system alone cost over 1/3 of a million dollars and it used several different GPS systems to cross check the fiber gyro.
The only way cars are going to take over for driving the mini-van in place of the drive soccer mom is if there is a serious attempt to clean up the road markings. This means no more optional parking on the side as a road will either be a parking spot or a lane. Signs will need to be redone and cleaned up. The white lines must be far more precise than they are now and more places will need to deal with the yellow centerline (which has now been dropped in the EU even though its the cheapest road safety device ever)
Things have gone a long way. 2 decades ago I had a system that would indicate that a steering adjustment needed to be done. A decade ago there was Miata convertible that could maintain road position and deal with deer. This year we have a VW that can avoid traffic cones. Maybe in a decade we can see a car that can avoid the phone talking, breakfast eating SUV driver.
Most people can't cope with the n'th letter at all. If they can cope with it, then their password entry will be so slow that shoulder surfing will be more than enough. There are a few very well documented psychological reasons why this is true. Try typical social engineering techniques before you recommend them. A small amount of experimentation goes a very long way.
Your getting into complexities that I believe won't help. Go to your CD collection and get the most obscure thing you have where you have at least 3 CD's and do the experiment. Then google for the lines you used. Your mind will be slightly tainted towards the most obscure lines you can think of because of this discussion but I expect you'll be surprised with the results.
The not disclosing the song is a given but its irrelevant since external hackers just use dictionaries anyway. I've seen dictionaries with song lyrics and Shakespeare over a decade ago.
Find someone who has a few CDs from one of the artists that you have the same albums. Have each of you write down three lines out from each song on each album and then compare lists. The statistics of this end up being very much like the birthday paradox in that as you add more people to the pool, the collision domain shrinks to nothing very quickly.
The scary thing about "password best practice" is that its all based on something that sounded like a good idea at the time (and might have been). I have friends who happen to like obscure songs written a thousand years ago in Latin but if they did the experiment above, they will come up with about the same results. Add in a random group of musicians in the SCA and the chance of an obscure song goes to zero.
How many memorable lines are their out of top 40 songs over the last 40 years? I'll give you a hint, its much smaller than you think. If there was one memorable line out of every top 40 song and the list was completely replaced every week, it would be about 4 times bigger than the number of words in a grade school student dictionary. Of course some songs have maybe a dozen lines that their fans will remember but some songs have been in the top 40 for a very long time and a busy month may have nearly a 1/4 of the list replaced.
What GMS phones do is cause spikes in the thing in the front of the plane that tells the pilot when they are over the VOR station and should do their turn. If the plane is in the clouds where they have rocks in the clouds and your pilot thinks they are over the VOR and turns into the rocks, you won't be around to tell people that phones don't cause any problems.
Phones aren't allowed to be used at the start of the flight when the VOR receiver is being calibrated either.
Plus the outer marker radios pick up that SMS pulse noise thats typical when your phone is near speakers and no one wants to listen to that crud.
If you don't like the seat pitch, cacluate how much you pay for that in economy vs business and you'll be even less happy. Business class tends to have 4/3 the pitch and 3/2 the width for 2x the total space yet business class tickets aren't 2x as expensive. It gets worse with 1st class with its pitch being 6/3 to 8/3 of cattle class.
If I have a disk with a bad bearing or head crash, I have to send the drive back intact to claim my warranty and so I'm trusting my data to someone else. At work thats a violation of our security policy so we don't ever send disks back unless we are very sure what data was on it which effectively means drives that died within a few hours of being installed in a machine.
What needs to happen is the S.M.A.R.T. software on the drive needs to be a bit smarter and allow the diagnostics tools to coordinate with the manufacturers web site so that the company is sure the disk is in fact defective and they know the cause. If they determine that they don't want the disk back anyway, then let the consumer dispose of it. If they can repair the disk, then its most likely not an internal problem so if its repairable, the data should still be on it when it gets back.
A few years ago a radio station in Melbourne ran a contest where bands sent in CD's that they had made within the year. The result was from a listening population area of about 3 million people, they got about 3000 CD each with a normal amount of songs (about 10 I'm guessing). That was a few years ago and before the advent of a real cheap way to make good audio CDs in your basement but if the figures hold true, I expect that there are about about 1000 bands who can make a CD per million people in most developed countries. Since there are nearly a billion people who fit into that group, I expect that there are about 10 million new songs recorded every year. On a small fraction of them ever get to the local record store.
Details of your hashed root password are leaked by several operating systems while pretending to secure against packet sequence numbers. Since all they are doing is hashing a hashed password and a some other guessable things that could be pre-computed or queried, your roots security have already been compromised.
I have a small program that mmaps a bit of code and then points the program counter at it. Everything runs fun until a OS call happens. I've heard that Mach allows user land programs to install their own OS calls but I haven't seen any example code to do it and I suspect such a feature isn't in OS X. I've hunted through the source and I while I could write my own system call and compile it in, there should be an easier way. Can anyone point me in the correct direction?
SMF doesn't monitor sub task any more than init does. If a sub package dies while the monitoring thing is watching and being overloaded with signals, then it gets missed with either system. Try it... except that init seems to be far more robust getting SIG_TERM signals from odd processes. Go find the sql databases that SMF use and start poking around with a binary editor (hacker style) and see what happens. Add a null into the shutdown record and see what happens when you run the shutdown command. Now how can you audit for that situation? If someone hacks your inittab, its going to show up unless they also hack all your standard tools. Same with RC scripts. I don't like systems that hide so much behind the scenes that I can't tell exactly what's going on.
I've been told the new stuff is there for my own good. SSHd can now depend on having a running name server... oh thats cool but I guess it means that if named dies at boot I can't get into the box can I? I like the old sequential system since I was in control and I knew the order things happened. If I wanted something to happen in parallel I had this cute trick with the magic '&' but I guess that was too hard and now I've got 800x more unauditable code to hope just works and few hours of poking and prodding shows its not up to the standards of Solaris 9. You may call that FUD. I just call it uncertainty, and doubt. I have no fear since this junk isn't going on my secure network.
Your comment about what happens in a zone stays there when I gave you an example that disproves it. Do a package in a zone and kill it. Then go into the global zone and try to start up pkgadd. You can't because something in a zone locks it out. Thats a serious design flaw that I would have expected someone to test and fix before shipping. There also appears to be some problem with doors between zones that I haven't had time to look into yet.
That link you provided claims jail setup requires "Yes, requires complex scripts". Isn't live update just a huge mess of complex scripts? I still haven't seen a way to install just a base install in a zone without pulling in all the tools that a cracker would use against me. My bind jails don't even have a shell in them. That can not be done with zones. The only thing solaris does better is resource control but others are catching up.
Maybe they found out that its a business requirement for many companies that have a security policy. We don't know what happens to a hard disk and the data that was on it when its sent back for warranty repairs so we take them out before sending back computers.
Check out the "Innocent third party" laws in your state. When my server got targeted by Sween, I sent a nasty letter to Microsoft and they took care of the bandwidth costs. They know they are liable and the law agrees, I just wonder why some lawyer has used these laws to take some of Billy Gates money away.
They have no focus and they have no direction and its costing them customers. My problem with them is they are throwing away their roots to be buzzword compliant and the making up buzzwords to meet that goal. I've been running sun boxes since the sun 3 days and like their hardware and I had a great respect for their kernel. Then with Soalris 10 they go and add a bunch off new features that I have no need for that break by security auditing procedures. They break core bits of the Unix concept such as init. Their new init requires 800 times more code and libraries than the init for Solaris version 7 and it fills my logs with crud if I choose to remove the smf junk out of initab. Guys its Unix, If I tell it to do something, I want it to do it.
The zones stuff looks cool but it hasn't been tested. If you lock up a pkgadd in a zone, then the root zone can't install patches. What happens in a zone should stay in a zone. I'm getting the feeling that doors cross zones with a bit too much ease. I still don't know why I need a complete operating environment in a zone. I like by BSD jails with just enough stuff in them to start the daemon and thats it.
I still don't know why sun thinks I need a word processor in my base install for a headless server.
We will be a Solaris shop until we can't get Solaris 9 anymore. We are not moving to Solaris 10 and we won't be buying anymore hardware that requires Solaris 10.
The T1000 needs serious tuning to not be a bad machine under every load I've tried. Its 39 times slower than a 1.25 GHz mini mac doing a single floating point process. Its integer performance per core is on par with a 500 MHz G4 PPC. Once its CPU bound, its a real dog but tuning can help some. Under heavy load its slower than a decade old SS1000. Its RSA engine is fast but its integer performance is so bad that for a single connection, the mini-mac is faster after 3k of compressed SSL (RSA+RC4) data. The only thing I have seen that the T1000 is faster at is poorly written multi-threaded java apps. It may work well for programs that are just a mess of indirection to indirection to indirection that are multi process and multi thread. My tests seem to show that a PPC Xserve cluster node should be faster than the T1000 for the loads that the T1 is best at and much faster for everything else for about the same amount of cash. The loads I looked at were MRTG, email, apache (w & w/o ssl), perl cgi programs. The boss won't let me do an aerodynamic benchmark of the T1000.
This has happened to many large groups that used SSN as a key. The SSA claimed as late as the 1987 that a SSN is not unique but it will be unique with for a given name however they would issue a new ssn if the last name and 1st initial were also the same.
I think the easy solution for this is for them to start printing a few extra random digits on about 5% of the next few batches of cards. How many systems would break if people started showing up with 14 digit SSNs
The only fix for open source patent issues is to make sure that stock holders get stung but the requires having a unified way to remove licenses from stock holders. An example could be that you end up losing the right to use Apache if you owned stock in a patent toll company.
Its easy to make a binary that is less subject to unknown attacks than the factory versions. I've been doing this for years and its not too hard. Start by building everything from source. Find the link order and change that around. Look at the build options since you may not need that -O2. There are programs that will rearrange the order of the variables which changes the stack order and some will even rearrange the calling order. You can even add filler as well. If your going to rebuild an entire os, you could go far as to reorder the constants in/usr/include before you rebuild it. If a program expects syscall 4 to be open() and you've changed it to select(), their code is going to break real quick.
If I get rid of X, the zone stuff still works but zones require live update which require java which pulls in X which pulls in a window manager and all sorts of other crud.
As far as speed goes, I'm sure there are some things that are faster but I don't see any of them being close to the real world. From what I've seen if you get the T1 busy with any more than 2 jobs per core its game over.
When I was hanging around in computer stores in the lear 1980's the local sales guys didn't have any clue about what they were selling. There was only one computer store within a hour's drive where you could get real answers to real questions.
I just got a T1000 with the intent of replacing a decade old SparcServer 1000. I'm still hunting for a job that the T1000 is fast at. It appears each core is on par with a 500 MHz G4 PPC and multi-threading only helps for poorly optimized code when it can swap to another thread while it preloads other data. I expect that new design for an 8 core 500 MHz PPC chip would run very cool too.
Since the T1000 only runs Solaris 10, I've been looking into getting it slimed down so its suitable for living out in the DMZ. Everything is so intermixed now that you need a whole mess of stuff just to get the thing to boot. An example is init and its required libraries are now 800 times larger than the Solaris 7 version. If you look at the full package dependency list, you must have X for the zone tools. Now if you have 800 times more data just to get the thing to start its 1st processes, how much better does the MTBF have to be to have the same application MTBF?
Slashdot Mirror
As far as I know that requires a DPGS like system on the track with extra real time feedback to the car.
So they are cheating if you consider the real world.
I've been in a car that could drive its self on one very well surveyed road. If it got confused it would beep and assume the human was in control within a second. The internal guidance system alone cost over 1/3 of a million dollars and it used several different GPS systems to cross check the fiber gyro.
The only way cars are going to take over for driving the mini-van in place of the drive soccer mom is if there is a serious attempt to clean up the road markings. This means no more optional parking on the side as a road will either be a parking spot or a lane. Signs will need to be redone and cleaned up. The white lines must be far more precise than they are now and more places will need to deal with the yellow centerline (which has now been dropped in the EU even though its the cheapest road safety device ever)
Things have gone a long way. 2 decades ago I had a system that would indicate that a steering adjustment needed to be done. A decade ago there was Miata convertible that could maintain road position and deal with deer. This year we have a VW that can avoid traffic cones. Maybe in a decade we can see a car that can avoid the phone talking, breakfast eating SUV driver.
Most people can't cope with the n'th letter at all. If they can cope with it, then their password entry will be so slow that shoulder surfing will be more than enough. There are a few very well documented psychological reasons why this is true. Try typical social engineering techniques before you recommend them. A small amount of experimentation goes a very long way.
Your getting into complexities that I believe won't help. Go to your CD collection and get the most obscure thing you have where you have at least 3 CD's and do the experiment. Then google for the lines you used. Your mind will be slightly tainted towards the most obscure lines you can think of because of this discussion but I expect you'll be surprised with the results.
The not disclosing the song is a given but its irrelevant since external hackers just use dictionaries anyway. I've seen dictionaries with song lyrics and Shakespeare over a decade ago.
Find someone who has a few CDs from one of the artists that you have the same albums. Have each of you write down three lines out from each song on each album and then compare lists. The statistics of this end up being very much like the birthday paradox in that as you add more people to the pool, the collision domain shrinks to nothing very quickly.
The scary thing about "password best practice" is that its all based on something that sounded like a good idea at the time (and might have been). I have friends who happen to like obscure songs written a thousand years ago in Latin but if they did the experiment above, they will come up with about the same results. Add in a random group of musicians in the SCA and the chance of an obscure song goes to zero.
How many memorable lines are their out of top 40 songs over the last 40 years? I'll give you a hint, its much smaller than you think. If there was one memorable line out of every top 40 song and the list was completely replaced every week, it would be about 4 times bigger than the number of words in a grade school student dictionary. Of course some songs have maybe a dozen lines that their fans will remember but some songs have been in the top 40 for a very long time and a busy month may have nearly a 1/4 of the list replaced.
What GMS phones do is cause spikes in the thing in the front of the plane that tells the pilot when they are over the VOR station and should do their turn. If the plane is in the clouds where they have rocks in the clouds and your pilot thinks they are over the VOR and turns into the rocks, you won't be around to tell people that phones don't cause any problems.
Phones aren't allowed to be used at the start of the flight when the VOR receiver is being calibrated either.
Plus the outer marker radios pick up that SMS pulse noise thats typical when your phone is near speakers and no one wants to listen to that crud.
If you don't like the seat pitch, cacluate how much you pay for that in economy vs business and you'll be even less happy. Business class tends to have 4/3 the pitch and 3/2 the width for 2x the total space yet business class tickets aren't 2x as expensive. It gets worse with 1st class with its pitch being 6/3 to 8/3 of cattle class.
If I have a disk with a bad bearing or head crash, I have to send the drive back intact to claim my warranty and so I'm trusting my data to someone else. At work thats a violation of our security policy so we don't ever send disks back unless we are very sure what data was on it which effectively means drives that died within a few hours of being installed in a machine.
What needs to happen is the S.M.A.R.T. software on the drive needs to be a bit smarter and allow the diagnostics tools to coordinate with the manufacturers web site so that the company is sure the disk is in fact defective and they know the cause. If they determine that they don't want the disk back anyway, then let the consumer dispose of it. If they can repair the disk, then its most likely not an internal problem so if its repairable, the data should still be on it when it gets back.
A few years ago a radio station in Melbourne ran a contest where bands sent in CD's that they had made within the year. The result was from a listening population area of about 3 million people, they got about 3000 CD each with a normal amount of songs (about 10 I'm guessing). That was a few years ago and before the advent of a real cheap way to make good audio CDs in your basement but if the figures hold true, I expect that there are about about 1000 bands who can make a CD per million people in most developed countries. Since there are nearly a billion people who fit into that group, I expect that there are about 10 million new songs recorded every year. On a small fraction of them ever get to the local record store.
Details of your hashed root password are leaked by several operating systems while pretending to secure against packet sequence numbers. Since all they are doing is hashing a hashed password and a some other guessable things that could be pre-computed or queried, your roots security have already been compromised.
Thanks for that link. It wasn't quite what I was looking for but it linked to some very interesting things that may lead me in the right direction.
I have a small program that mmaps a bit of code and then points the program counter at it. Everything runs fun until a OS call happens. I've heard that Mach allows user land programs to install their own OS calls but I haven't seen any example code to do it and I suspect such a feature isn't in OS X. I've hunted through the source and I while I could write my own system call and compile it in, there should be an easier way. Can anyone point me in the correct direction?
SMF doesn't monitor sub task any more than init does. If a sub package dies while the monitoring thing is watching and being overloaded with signals, then it gets missed with either system. Try it... except that init seems to be far more robust getting SIG_TERM signals from odd processes. Go find the sql databases that SMF use and start poking around with a binary editor (hacker style) and see what happens. Add a null into the shutdown record and see what happens when you run the shutdown command. Now how can you audit for that situation? If someone hacks your inittab, its going to show up unless they also hack all your standard tools. Same with RC scripts. I don't like systems that hide so much behind the scenes that I can't tell exactly what's going on.
I've been told the new stuff is there for my own good. SSHd can now depend on having a running name server... oh thats cool but I guess it means that if named dies at boot I can't get into the box can I? I like the old sequential system since I was in control and I knew the order things happened. If I wanted something to happen in parallel I had this cute trick with the magic '&' but I guess that was too hard and now I've got 800x more unauditable code to hope just works and few hours of poking and prodding shows its not up to the standards of Solaris 9. You may call that FUD. I just call it uncertainty, and doubt. I have no fear since this junk isn't going on my secure network.
Your comment about what happens in a zone stays there when I gave you an example that disproves it. Do a package in a zone and kill it. Then go into the global zone and try to start up pkgadd. You can't because something in a zone locks it out. Thats a serious design flaw that I would have expected someone to test and fix before shipping. There also appears to be some problem with doors between zones that I haven't had time to look into yet.
That link you provided claims jail setup requires "Yes, requires complex scripts". Isn't live update just a huge mess of complex scripts? I still haven't seen a way to install just a base install in a zone without pulling in all the tools that a cracker would use against me. My bind jails don't even have a shell in them. That can not be done with zones. The only thing solaris does better is resource control but others are catching up.
Maybe they found out that its a business requirement for many companies that have a security policy. We don't know what happens to a hard disk and the data that was on it when its sent back for warranty repairs so we take them out before sending back computers.
Check out the "Innocent third party" laws in your state. When my server got targeted by Sween, I sent a nasty letter to Microsoft and they took care of the bandwidth costs. They know they are liable and the law agrees, I just wonder why some lawyer has used these laws to take some of Billy Gates money away.
They have no focus and they have no direction and its costing them customers. My problem with them is they are throwing away their roots to be buzzword compliant and the making up buzzwords to meet that goal. I've been running sun boxes since the sun 3 days and like their hardware and I had a great respect for their kernel. Then with Soalris 10 they go and add a bunch off new features that I have no need for that break by security auditing procedures. They break core bits of the Unix concept such as init. Their new init requires 800 times more code and libraries than the init for Solaris version 7 and it fills my logs with crud if I choose to remove the smf junk out of initab. Guys its Unix, If I tell it to do something, I want it to do it.
The zones stuff looks cool but it hasn't been tested. If you lock up a pkgadd in a zone, then the root zone can't install patches. What happens in a zone should stay in a zone. I'm getting the feeling that doors cross zones with a bit too much ease. I still don't know why I need a complete operating environment in a zone. I like by BSD jails with just enough stuff in them to start the daemon and thats it.
I still don't know why sun thinks I need a word processor in my base install for a headless server.
We will be a Solaris shop until we can't get Solaris 9 anymore. We are not moving to Solaris 10 and we won't be buying anymore hardware that requires Solaris 10.
The T1000 needs serious tuning to not be a bad machine under every load I've tried.
Its 39 times slower than a 1.25 GHz mini mac doing a single floating point process.
Its integer performance per core is on par with a 500 MHz G4 PPC.
Once its CPU bound, its a real dog but tuning can help some. Under heavy load its slower than a decade old SS1000.
Its RSA engine is fast but its integer performance is so bad that for a single connection, the mini-mac
is faster after 3k of compressed SSL (RSA+RC4) data.
The only thing I have seen that the T1000 is faster at is poorly written multi-threaded java apps. It may work well for programs that are just a mess of indirection to indirection to indirection that are multi process and multi thread.
My tests seem to show that a PPC Xserve cluster node should be faster than the T1000 for the loads that the T1 is best at and much faster for everything else for about the same amount of cash. The loads I looked at were MRTG, email, apache (w & w/o ssl), perl cgi programs.
The boss won't let me do an aerodynamic benchmark of the T1000.
Apple drives are slow unless you buy their media. Who knows how long their burner scams will continue.
This has happened to many large groups that used SSN as a key. The SSA claimed as late as the 1987 that a SSN is not unique but it will be unique with for a given name however they would issue a new ssn if the last name and 1st initial were also the same.
I think the easy solution for this is for them to start printing a few extra random digits on about 5% of the next few batches of cards. How many systems would break if people started showing up with 14 digit SSNs
The only fix for open source patent issues is to make sure that stock holders get stung but the requires having a unified way to remove licenses from stock holders. An example could be that you end up losing the right to use Apache if you owned stock in a patent toll company.
Its easy to make a binary that is less subject to unknown attacks than the factory versions. I've been doing this for years and its not too hard. Start by building everything from source. Find the link order and change that around. Look at the build options since you may not need that -O2. There are programs that will rearrange the order of the variables which changes the stack order and some will even rearrange the calling order. You can even add filler as well. If your going to rebuild an entire os, you could go far as to reorder the constants in /usr/include before you rebuild it. If a program expects syscall 4 to be open() and you've changed it to select(), their code is going to break real quick.
If I get rid of X, the zone stuff still works but zones require live update which require java which pulls in X which pulls in a window manager and all sorts of other crud.
As far as speed goes, I'm sure there are some things that are faster but I don't see any of them being close to the real world. From what I've seen if you get the T1 busy with any more than 2 jobs per core its game over.
And its bigger than my 8310 which they no longer make.
When I was hanging around in computer stores in the lear 1980's the local sales guys didn't have any clue about what they were selling. There was only one computer store within a hour's drive where you could get real answers to real questions.
I just got a T1000 with the intent of replacing a decade old SparcServer 1000. I'm still hunting for a job that the T1000 is fast at. It appears each core is on par with a 500 MHz G4 PPC and multi-threading only helps for poorly optimized code when it can swap to another thread while it preloads other data. I expect that new design for an 8 core 500 MHz PPC chip would run very cool too.
Since the T1000 only runs Solaris 10, I've been looking into getting it slimed down so its suitable for living out in the DMZ. Everything is so intermixed now that you need a whole mess of stuff just to get the thing to boot. An example is init and its required libraries are now 800 times larger than the Solaris 7 version. If you look at the full package dependency list, you must have X for the zone tools. Now if you have 800 times more data just to get the thing to start its 1st processes, how much better does the MTBF have to be to have the same application MTBF?