why would that configuration be insanity. It gives you a local cache while still letting your ISP (which has a bigger pool of requests to work with and a faster connection) do it's own caching and the resolving work.
you shouldn't need to shut down the grid to resync, just pump in slightly more or slightly less power until they line up then throw the switch. Once they are connectected to the grid generators will stay in sync by themselves.
Hell, there's a dozen guys here on slashdot who think they can just dash off a file system any time they feel like it. Writing a filesystem and writing a good filesystem are different things.
A basic filesystem similar to FAT could be designed and coded by any decent coder in a fairly short period of time.
A filesystem that is competitive in terms of performance and reliability with modern filesystems is much much harder.
What about the cost of keeping the guilty person alive and imprisoned for the rest of their life? Last I heared in the USA it cost more to get someone executed than to keep them in prision for life.
Afaict the main reason for the death penalty in the US is so they can plea bargin with people and yet still keep them out of society forever.
AC-to-DC conversion is messy and lossy. DC-DC voltage conversion isn't exactly great either, particularlly if you want it isolated.
Distributing at the final utilisation voltage has two problems. Firstly many systems need a wide range of voltages. Secondly long cables at 12V or 5V or worse 3.3V are also very lossy.
Distributing at 48V DC or so means an extra conversion step. There are also issues with safely switching DC.
For datacenters and other UPS supplied systems DC distribution usually at 48V or so can make sense because you can skip out a conversion step and because it's much easier to feed your protected distribution network from multiple sources.
Blackboard locks down information There seems to be a certain posessiveness about lecture notes among many academics. I guess it's not surprising when you hear stories of chineese universities copying entire sets of lecture notes from their western counterparts and then to add insult to injury claiming them as thier own work.
As a way of marking up text with formatting and structural information html and similar xml schemas rock. But if you try to use them for other things you tend to end up with more markup than text and a file that is almost unreadable.
One possibility would be to use XML for handling the tree but store the bottom level values as name=value pairs in the body text. That way you could leverage an xml parser for most of the work while giving a far more readable file (similar to the apache configuration file).
It really depends what you count as embedded. Some vendors already offer nas soloutions based on windows server 2003. Of course the only real difference between theese and more conventional low end servers is the admin tools and the licensing model.
Running on systems that aren't based on x86,x64 or itanium seems unlikely.
cyrix/IDT/via - cheap procs, now trying to push micro form factors. I was under the impression that the epia boards were mainly made the embedded/thin client market. Some geeks bought them for small form factor PCs but they are really too far behind to be good in that role. As you say the majority of micro form factor machines now are using intel laptop chips which aren't much bulkier and give much much better performance.
making up a call seems like a horriblly bad idea. Law enforcement would probablly go over your phone records with a fine tooth comb after that.
Re:It is all about the platform.
on
Is AMD Dead Yet?
·
· Score: 1
P2 -> P3, new chipset, new socket, new motherboard Early P3's used slot 1 just like P2's did. Intel later introduced the P3 in socket 370 form factor and eventually phased out slot 1 but thier were third party adaptors availible and afaict even the last P3's could be used in a BX based board with the right adaptor.
As for your AMD information that looks plain wrong. I'm 99% sure the 64 bit chips all needed new sockets and both the XP and the thunderbird athlon and the duron were on socket A (which is missing for your list). You are also missing slot A (pre-thunderbird athlon) and socket 7 off the beggining of your AMD list (which were both used for AMD processors during the intel PII era.
I agree intel has been going through new socket types at an appalling rate recently but AMD has been no saint in this regard either.
I cannot imagine that BGP allows IP-address hijacking BGP allows routers to negotiate the best path to a destination, to do this peers must accept route information from each other the level of trust placed in a router by another router depends on the configuration.
If you are a small customer your upstream is likely to put filters on you to make sure you can't announce prefixes you don't own.
If you are a major ISP doing that would start to get impractical. Many customers will buy thier service off you and customers are likely to come and go on a fairly regular basis. Having to get your upstream and peers to manually unblock every route you announce would be a major PITA and unless they actively checked everyone (completely impracical) it wouldn't help much anyway. So your upstream just trusts you.
Re:Free Software friendly
on
Is AMD Dead Yet?
·
· Score: 2, Interesting
I looked through the first and last page of that thread and didn't find anything relavent, do you have a better link than a 13 page! forum thread that might have the information burried in the middle somewhere.
Re:x86 history report...
on
Is AMD Dead Yet?
·
· Score: 2, Interesting
I remember back in the 486/pentium era there were at least three smaller vendors competing with intel for the low end PC market (IDT, cyrix and AMD). AMD are still competitive, IDT and cyrix essentially failed in the general purpose PC cpu market and thier PC processor related stuff ended up owned by VIA who use thier IP to produce low performance low power PC compatible processors for embedded and thin client markets but they don't even try to compete on price/performance even at the low end of the market.
Since then afaict there has been on new entry (transmeta) in the PC processor market which was a miserable failure.
So we are down from three to one serious competitors for intel. If AMD fall I wonder if anyone else will ever manage to break in.
People don't care, because it'll be easy to port their apps anyway. Afaict it is pretty hard to port an app to pure.net (with no calls at all into custom native code), certainly much more trouble than porting to a new CPU architecture.
Besides look how long win16 has hung arround, killing off win32 will be suicidal for a long time even if porting an app you have the source for is trivial. MS lives on the fact that binaries for the most part just keep working.
Re:Why did they buy ATI?
on
Is AMD Dead Yet?
·
· Score: 4, Interesting
When you are designing architectures for 7 or so years out, you need a powerful crystal ball, but no such thing exists. AMD just guessed wrong about the nature of future applications. Intel guessed wrong with the Itanium also. Maybe the common thread is you have to fit existing apps instead of the other way around. But, betting against app change has risk also.
The problem is that the low end is probablly only a couple of years behind the high end. So if you try and stick to the low end you still have to design architectures 5 years or so out and each low end CPU makes far less profit than each high end CPU so you find it even harder to cover those R&D costs.
Any costs incurred by Microsoft will surely be passed along to their customers, Surely MS would already be pricing windows at the price point that they think will make them the most profit.
One big problem with class actions afaict is that they often end up with a settlement where the laywers get paid real money but all the victims get is some near worthless vouchers.
IMO class actions should never be settled, any restitution should be paid with no strings attatched and the money for victims who have not claimed yet should be held safely for those victims until they do.
The U.S. court system is currently biased against punitive damages because often even when the defendant deserves to have to pay, the victim doesn't deserve the money so the court system errs on the side of the defendant. This change would help fix that. Personally I think civil courts should not be allowed to get involved with punative matters. I find it crazy that you can ruin a person or company financially through punitive damages with only "balance of probabilities" evidence. If you wan't to punish people then you should use criminal court and fines.
Drivers running in user mode How many drivers really run in user mode though? The only hardware category i'm aware of where MS forced vendors to go user mode where they didn't have to before is the rendering component of printer drivers.
One big problem is that once two or more groups of people start hating and attacking each other it is very very difficult to break the cycle. They hate and/or feel threatened by each other so they attack each other. This makes them hate each other and feel threatened by each other even more.
in general stability and prosperity breeds more stability and prosperity while instability and poverty breeds more instability and poverty. Unfortunately it is very hard to make a country stable and prosperous against their will, not to mention not nessacerally in our (as westerners) best interests.
afaict the worst they could do is force them to start buying system builder packs like the whitebox vendors do. This is probablly more expensive than what the big brands pay but given the price of sonys laptops it probablly wouldn't be a significant part of the price. It would be a major incoviniance though having to enter a seperate key for every machine and activate it online.
the current development restrictions that in practice gain Sony nothing Games consoles are generally sold on a similar principle to inkjets and cartridges or razors and blades. The console is sold cheap (sometimes at a loss sometimes at a small profit) and the real money is made from the games.
That afaict is why the linux system is locked out of the 3D graphics. If it wasn't locked out people would be able to develop and market games without going through the official channels. That would be bad for the game revenue stream.
debian testing is not that unstable really, you wouldn't wan't to use it for really critical stuff but I really can't see how having a couple of your more savvy users on it so you notice any problems before they become serious on it would be a problem. I can't really comment on other distros.
If you don't do your testing while the distro is still in open development it gets much much harder to get any problems you find fixed since (understandablly) any changes during release freezes must go through far more beuracracy and may be outright rejected if the risk is deemed to outweigh the benifit.
Slashdot Mirror
why would that configuration be insanity. It gives you a local cache while still letting your ISP (which has a bigger pool of requests to work with and a faster connection) do it's own caching and the resolving work.
you shouldn't need to shut down the grid to resync, just pump in slightly more or slightly less power until they line up then throw the switch. Once they are connectected to the grid generators will stay in sync by themselves.
http://web.archive.org/web/20021130135854/http://politicalhumor.com/HumorPics/NoTalibanAnsMach.mp3
Hell, there's a dozen guys here on slashdot who think they can just dash off a file system any time they feel like it.
Writing a filesystem and writing a good filesystem are different things.
A basic filesystem similar to FAT could be designed and coded by any decent coder in a fairly short period of time.
A filesystem that is competitive in terms of performance and reliability with modern filesystems is much much harder.
What about the cost of keeping the guilty person alive and imprisoned for the rest of their life?
Last I heared in the USA it cost more to get someone executed than to keep them in prision for life.
Afaict the main reason for the death penalty in the US is so they can plea bargin with people and yet still keep them out of society forever.
AC-to-DC conversion is messy and lossy.
DC-DC voltage conversion isn't exactly great either, particularlly if you want it isolated.
Distributing at the final utilisation voltage has two problems. Firstly many systems need a wide range of voltages. Secondly long cables at 12V or 5V or worse 3.3V are also very lossy.
Distributing at 48V DC or so means an extra conversion step. There are also issues with safely switching DC.
For datacenters and other UPS supplied systems DC distribution usually at 48V or so can make sense because you can skip out a conversion step and because it's much easier to feed your protected distribution network from multiple sources.
Blackboard locks down information
There seems to be a certain posessiveness about lecture notes among many academics. I guess it's not surprising when you hear stories of chineese universities copying entire sets of lecture notes from their western counterparts and then to add insult to injury claiming them as thier own work.
As a way of marking up text with formatting and structural information html and similar xml schemas rock. But if you try to use them for other things you tend to end up with more markup than text and a file that is almost unreadable.
One possibility would be to use XML for handling the tree but store the bottom level values as name=value pairs in the body text. That way you could leverage an xml parser for most of the work while giving a far more readable file (similar to the apache configuration file).
It really depends what you count as embedded. Some vendors already offer nas soloutions based on windows server 2003. Of course the only real difference between theese and more conventional low end servers is the admin tools and the licensing model.
Running on systems that aren't based on x86,x64 or itanium seems unlikely.
cyrix/IDT/via - cheap procs, now trying to push micro form factors.
I was under the impression that the epia boards were mainly made the embedded/thin client market. Some geeks bought them for small form factor PCs but they are really too far behind to be good in that role. As you say the majority of micro form factor machines now are using intel laptop chips which aren't much bulkier and give much much better performance.
making up a call seems like a horriblly bad idea. Law enforcement would probablly go over your phone records with a fine tooth comb after that.
P2 -> P3, new chipset, new socket, new motherboard
Early P3's used slot 1 just like P2's did. Intel later introduced the P3 in socket 370 form factor and eventually phased out slot 1 but thier were third party adaptors availible and afaict even the last P3's could be used in a BX based board with the right adaptor.
As for your AMD information that looks plain wrong. I'm 99% sure the 64 bit chips all needed new sockets and both the XP and the thunderbird athlon and the duron were on socket A (which is missing for your list). You are also missing slot A (pre-thunderbird athlon) and socket 7 off the beggining of your AMD list (which were both used for AMD processors during the intel PII era.
I agree intel has been going through new socket types at an appalling rate recently but AMD has been no saint in this regard either.
I cannot imagine that BGP allows IP-address hijacking
BGP allows routers to negotiate the best path to a destination, to do this peers must accept route information from each other the level of trust placed in a router by another router depends on the configuration.
If you are a small customer your upstream is likely to put filters on you to make sure you can't announce prefixes you don't own.
If you are a major ISP doing that would start to get impractical. Many customers will buy thier service off you and customers are likely to come and go on a fairly regular basis. Having to get your upstream and peers to manually unblock every route you announce would be a major PITA and unless they actively checked everyone (completely impracical) it wouldn't help much anyway. So your upstream just trusts you.
I looked through the first and last page of that thread and didn't find anything relavent, do you have a better link than a 13 page! forum thread that might have the information burried in the middle somewhere.
I remember back in the 486/pentium era there were at least three smaller vendors competing with intel for the low end PC market (IDT, cyrix and AMD). AMD are still competitive, IDT and cyrix essentially failed in the general purpose PC cpu market and thier PC processor related stuff ended up owned by VIA who use thier IP to produce low performance low power PC compatible processors for embedded and thin client markets but they don't even try to compete on price/performance even at the low end of the market.
Since then afaict there has been on new entry (transmeta) in the PC processor market which was a miserable failure.
So we are down from three to one serious competitors for intel. If AMD fall I wonder if anyone else will ever manage to break in.
People don't care, because it'll be easy to port their apps anyway. .net (with no calls at all into custom native code), certainly much more trouble than porting to a new CPU architecture.
Afaict it is pretty hard to port an app to pure
Besides look how long win16 has hung arround, killing off win32 will be suicidal for a long time even if porting an app you have the source for is trivial. MS lives on the fact that binaries for the most part just keep working.
When you are designing architectures for 7 or so years out, you need a powerful crystal ball, but no such thing exists. AMD just guessed wrong about the nature of future applications. Intel guessed wrong with the Itanium also. Maybe the common thread is you have to fit existing apps instead of the other way around. But, betting against app change has risk also.
The problem is that the low end is probablly only a couple of years behind the high end. So if you try and stick to the low end you still have to design architectures 5 years or so out and each low end CPU makes far less profit than each high end CPU so you find it even harder to cover those R&D costs.
Any costs incurred by Microsoft will surely be passed along to their customers,
Surely MS would already be pricing windows at the price point that they think will make them the most profit.
One big problem with class actions afaict is that they often end up with a settlement where the laywers get paid real money but all the victims get is some near worthless vouchers.
IMO class actions should never be settled, any restitution should be paid with no strings attatched and the money for victims who have not claimed yet should be held safely for those victims until they do.
The U.S. court system is currently biased against punitive damages because often even when the defendant deserves to have to pay, the victim doesn't deserve the money so the court system errs on the side of the defendant. This change would help fix that.
Personally I think civil courts should not be allowed to get involved with punative matters. I find it crazy that you can ruin a person or company financially through punitive damages with only "balance of probabilities" evidence. If you wan't to punish people then you should use criminal court and fines.
Drivers running in user mode
How many drivers really run in user mode though? The only hardware category i'm aware of where MS forced vendors to go user mode where they didn't have to before is the rendering component of printer drivers.
One big problem is that once two or more groups of people start hating and attacking each other it is very very difficult to break the cycle. They hate and/or feel threatened by each other so they attack each other. This makes them hate each other and feel threatened by each other even more.
in general stability and prosperity breeds more stability and prosperity while instability and poverty breeds more instability and poverty. Unfortunately it is very hard to make a country stable and prosperous against their will, not to mention not nessacerally in our (as westerners) best interests.
afaict the worst they could do is force them to start buying system builder packs like the whitebox vendors do. This is probablly more expensive than what the big brands pay but given the price of sonys laptops it probablly wouldn't be a significant part of the price. It would be a major incoviniance though having to enter a seperate key for every machine and activate it online.
the current development restrictions that in practice gain Sony nothing
Games consoles are generally sold on a similar principle to inkjets and cartridges or razors and blades. The console is sold cheap (sometimes at a loss sometimes at a small profit) and the real money is made from the games.
That afaict is why the linux system is locked out of the 3D graphics. If it wasn't locked out people would be able to develop and market games without going through the official channels. That would be bad for the game revenue stream.
debian testing is not that unstable really, you wouldn't wan't to use it for really critical stuff but I really can't see how having a couple of your more savvy users on it so you notice any problems before they become serious on it would be a problem. I can't really comment on other distros.
If you don't do your testing while the distro is still in open development it gets much much harder to get any problems you find fixed since (understandablly) any changes during release freezes must go through far more beuracracy and may be outright rejected if the risk is deemed to outweigh the benifit.
you can detect a bug like that on your POTS line by monitoring the voltage on the line.
I call bullshit
provided the bug isn't trying to draw power from the line and has a nice high impedance input stage I very much doubt you could reliablly detect it.