can't say i've ever had trouble with USB on 2K, as for CD burning iirc the stuff shipped with windows is pretty crippled though it may be handy in a pinch.
"stop and ask "wait a minute, why do we do these upgrades again?"." In a large networked environment basing your desktops on a version of windows that no longer gets security updates does not seem sensible to me. Also as a release gets older finding suitable hardware/software/drivers for it will get harder and harder.
So you either have to let new machines have the newest as they come in (meaning you support a mixture of releases) or you do a mass upgrade at some point in the release cycle.
ok several portable DVD players was probablly going over the top but some richer families probablly have a portable, players in the car and so on as well as players hooked up to a TV.
HDTV penetration was only 25% in the US. And what defintion of penatration is that using?
I bet there are a lot of households with one HDTV but a whole load of non HDTVs several of which will have DVD players hooked up to them. And probablly several portable DVD players as well. All those DVD players are going to take a long time to cycle out of the system.
Is it really in the movie companies interests for blue-ray to replace DVD? I would have thought it would be more profitable to keep selling DVDs to normal customers at reasonable prices and use blue-ray to milk the movie buffs.
But from a consumers standpoint, Blu-Ray has less DRM - because there are fewer regions. The biggest one is that we are in the same region as Japan. ok so americans with a fetish for japaneese anime may well preffer blue-ray but I doubt they are a significant part of the market.
And at least here in europe multi-region DVD players are pretty easy to get hold of but i've never seen a multi-region blue-ray player advertised.
And then there is HDCP which can be a major PITA in some setups.
And then there are all the people who use DVD copy software that relys on the fact that DVD drm was pretty thouroughly cracked some time ago (blue-ray DRM is still an arms race afaict)
The thing about ram is that it is defect tollerant. It's easy to add a bit of extra ram and make the decode circuits programable. Then after testing the chip they can program it to not use the defective blocks.
With less homogenous chips like CPUs this is much harder.
A 13 year old linux install would mean something released before debian 1.1 (due to a release fuckup there was never a debian 1.0) and probablly before the elf transition. Have fun upgrading that though the many releases to get to something reaonablly current. I can be pretty sure that at least some of those upgrades (plural because afaict linux distributions tend not to support upgrading more than one release at a time) will involve the loss of some software and/or some manual intervention.
Good for you, unfortunately many people when given the ability to spend way beyond thier means by the dodgy mortgage brokers did so. Of course in many areas housing is in limited supply so more money availible to borrow just means prices go up meaning even more people end up taking out loans that they can't really afford.
Now there has been a massive bust and the feds have to work out how to deal with it without destroying the rest of the US economy and possiblly several other countries economies too.
I dunno what it is like in the US but here in the UK a lot of people tend to vote primerally based on the party not the individual.
What that means is that if a member of parliment pisses off thier party the party can kick them out. Once kicked out of the party they will find it very hard to hold on to thier seat in parliment since as well as the oposition party they will also be competing against thier old parties official candidate.
Your apps should be relying on the OS to handle the nastiness of networking. Maybe they should but the fact is that details on the addressing scheme are exposed to applications.
A lot of sockets using apps are written with the assumption that they will only ever be used on IPV4 sockets. IPV6 requires larger sockaddr structures and new name/address conversion routines. There can often be other complications too like configuration files that use : as a field seperator.
If it's only behind a nat that you have configuration access to it's trivial, just configure the nat to forward ports appropriately.
On the other hand if it's behind a nat someone else (e.g. your ISP) owns and doesn't let you have configuration access to then there is more of a problem.
When talking about consumer equipment NAT almost always reffers to a one to many NAT. A one to many NAT has to be doing state tracking to know where to send the packets and unless there is a fallback rule in place anything that doesn't match up with a mapping is going to get dropped or treated as a connection to the NAT box itself and rejected by it.
Generally your computer should not be connecting out from a port that is in use for a sever so that NAT should not be generating any mappings related to the port the service is running on.
The next big thing I imagine will be the reallocation of class A addresses: why should the likes of HP get multiple class A's? Afaict there is no law saying someone can't use any IP address they damn well please, it's just that if they wan't to participate in the public internet they have to follow the rules handed to them by thier internet provider.
What this means is that as long as ICANN act in a way the major ISPs consider tollerable they get to stay in charge. BUT if they were to try and revoke major allocations it could easilly cause some major ISPs to ignore them. Since the internet is reliant on addresses being globally unique this would be a disaster,
And with at least one major operating system refusing to use the class E space as normal space opening that up is not practical either.
An IP address market is an option but it would be likely to lead to a route table explosion unless the minimum resellable block is kept very large.
AFAIK there is only one real problem left that will keep many big businesses from deploying IPv6: multi-homing. The technology to have more than one upstream provider for IPv6 connectivity is still in flux. AFAICT for big buisnesses this is a non-issue, they will already be an "autonomous system" and will probablly have no trouble getting a globally routable/32 allocated which they can advertise on as many providers as they like.
For small and medium buisnesses there is a big problem though. With IPV4 you could use NAT and I belive you could also advertise space allocated by one provider on another provider. With ipv6 NAT is not readilly availible (linux doesn't support it for philosophical reasons and I don't think many other platforms support it either) and I don't think you are allowed to advertise space from one provider on another provider.
There was a system called A6 which was supposed to simplify handling of sites with multiple addresses running in paralell but it was extremely complicated in itself and has been abandoned afaict.
Just browsing through quickly, it seems that at least the extreme models ship with 4gb of RAM + video card memory. Assuming they use 32 bit window - due to the better drivers, from what I've seen it benchmarks better than 64 bit, isn't this completly pointless? under an OS that only supports 4GB of address space 4GB of ram tends to give you a little extra usable memory over 3GB (how much depends on the other hardware in the box). Also with some chipsets using a matched pair gives better performance than an unmatched pair.
It's probablly a marginal benefit but given the extremely low cost of ram now some may consider it worth having.
I guess it depends on what you consider a hardware upgrade. Huge numbers of games for older consoles had banking hardware and saveram (some of which may be used for things other than saves) in the cart and occasionally there was other stuff in there too.
DS carts on the other hand have an interface that is disk like rather than rom like so if you want extra ram you need to put it on the GBA slot. That means an extra cart for the user to lose. So far the only commercial DS product i'm aware of that does so is the web browser.
Using escaping to protect your queries is like doing strings by manually allocating blocks of memory and then storing a sequence of characters in them followed by a null terminator (either directly or through a series of helper functions).
Both can sometimes be forced on you by the environment you are working in. Both require an extreme level of attention to detail by both the initial programmer and later programmers who work on the code. Both can easilly lead to security holes if the programmer makes a simple mistake.
Houses have pretty big safety margins, if most of the house is built to code the odd bad beam or joint isn't going to bring things crashing down.
A driver is a peice of software written in a language without safety checks, running in an environment without process seperation, having to deal with obscure concepts like IRQL and often interacting directly with hardware. One bug can easilly bring the system crashing down and the chance of a test team finding every bug is pretty low even if they have access to read the source.
there's absolutely nothing performance critical enough to justify running it in kernel space. no but moving to a usermode driver would basically mean a complete rewrite and would reduce the number of windows versions supported. Doesn't seem very attractive to me.
The drivers are USB protocol drivers -- they run in user space. They can be written as userspace drivers but they probablly aren't. Moving your driver to userspace means rewriting it and losing compatibility with a number of older windows versions. Not a very attractive proposition IMO.
USB drivers live in userspace (only the generic read/write support for USB devices live in kernel space) Afaict on windows most USB drivers live in kernel space. It is possible to make user mode USB drivers but if you do afaict they will not work on 2K or server 2K3 (this may have changed but if it has MS didn't bother to update thier faq and I don't have access to download the wdk anymore). Furthermore a user mode driver can't have kernel mode clients (which means you can't do keyboard/mouse drivers or mass storage drivers in userspace)
Slashdot Mirror
can't say i've ever had trouble with USB on 2K, as for CD burning iirc the stuff shipped with windows is pretty crippled though it may be handy in a pinch.
"stop and ask "wait a minute, why do we do these upgrades again?"."
In a large networked environment basing your desktops on a version of windows that no longer gets security updates does not seem sensible to me. Also as a release gets older finding suitable hardware/software/drivers for it will get harder and harder.
So you either have to let new machines have the newest as they come in (meaning you support a mixture of releases) or you do a mass upgrade at some point in the release cycle.
ok several portable DVD players was probablly going over the top but some richer families probablly have a portable, players in the car and so on as well as players hooked up to a TV.
And players in thier computers too......
HDTV penetration was only 25% in the US.
And what defintion of penatration is that using?
I bet there are a lot of households with one HDTV but a whole load of non HDTVs several of which will have DVD players hooked up to them. And probablly several portable DVD players as well. All those DVD players are going to take a long time to cycle out of the system.
Is it really in the movie companies interests for blue-ray to replace DVD? I would have thought it would be more profitable to keep selling DVDs to normal customers at reasonable prices and use blue-ray to milk the movie buffs.
But from a consumers standpoint, Blu-Ray has less DRM - because there are fewer regions. The biggest one is that we are in the same region as Japan.
ok so americans with a fetish for japaneese anime may well preffer blue-ray but I doubt they are a significant part of the market.
And at least here in europe multi-region DVD players are pretty easy to get hold of but i've never seen a multi-region blue-ray player advertised.
And then there is HDCP which can be a major PITA in some setups.
And then there are all the people who use DVD copy software that relys on the fact that DVD drm was pretty thouroughly cracked some time ago (blue-ray DRM is still an arms race afaict)
Why not? Modern formats are variable bitrate anyway so lowering the average bitrate shouldn't cause any compatibility problems.
The thing about ram is that it is defect tollerant. It's easy to add a bit of extra ram and make the decode circuits programable. Then after testing the chip they can program it to not use the defective blocks.
With less homogenous chips like CPUs this is much harder.
I call bullshit
A 13 year old linux install would mean something released before debian 1.1 (due to a release fuckup there was never a debian 1.0) and probablly before the elf transition. Have fun upgrading that though the many releases to get to something reaonablly current. I can be pretty sure that at least some of those upgrades (plural because afaict linux distributions tend not to support upgrading more than one release at a time) will involve the loss of some software and/or some manual intervention.
Good for you, unfortunately many people when given the ability to spend way beyond thier means by the dodgy mortgage brokers did so. Of course in many areas housing is in limited supply so more money availible to borrow just means prices go up meaning even more people end up taking out loans that they can't really afford.
Now there has been a massive bust and the feds have to work out how to deal with it without destroying the rest of the US economy and possiblly several other countries economies too.
There is nothing of interest there.
Nothing of interest in one of the richest and most populus countries in the world? I highly doubt it.
I dunno what it is like in the US but here in the UK a lot of people tend to vote primerally based on the party not the individual.
What that means is that if a member of parliment pisses off thier party the party can kick them out. Once kicked out of the party they will find it very hard to hold on to thier seat in parliment since as well as the oposition party they will also be competing against thier old parties official candidate.
I strongly suspect the result of that would be the major ISPs telling ICANN to go screw themselves.
Your apps should be relying on the OS to handle the nastiness of networking.
Maybe they should but the fact is that details on the addressing scheme are exposed to applications.
A lot of sockets using apps are written with the assumption that they will only ever be used on IPV4 sockets. IPV6 requires larger sockaddr structures and new name/address conversion routines. There can often be other complications too like configuration files that use : as a field seperator.
If it's only behind a nat that you have configuration access to it's trivial, just configure the nat to forward ports appropriately.
On the other hand if it's behind a nat someone else (e.g. your ISP) owns and doesn't let you have configuration access to then there is more of a problem.
When talking about consumer equipment NAT almost always reffers to a one to many NAT. A one to many NAT has to be doing state tracking to know where to send the packets and unless there is a fallback rule in place anything that doesn't match up with a mapping is going to get dropped or treated as a connection to the NAT box itself and rejected by it.
Generally your computer should not be connecting out from a port that is in use for a sever so that NAT should not be generating any mappings related to the port the service is running on.
The next big thing I imagine will be the reallocation of class A addresses: why should the likes of HP get multiple class A's?
Afaict there is no law saying someone can't use any IP address they damn well please, it's just that if they wan't to participate in the public internet they have to follow the rules handed to them by thier internet provider.
What this means is that as long as ICANN act in a way the major ISPs consider tollerable they get to stay in charge. BUT if they were to try and revoke major allocations it could easilly cause some major ISPs to ignore them. Since the internet is reliant on addresses being globally unique this would be a disaster,
And with at least one major operating system refusing to use the class E space as normal space opening that up is not practical either.
An IP address market is an option but it would be likely to lead to a route table explosion unless the minimum resellable block is kept very large.
AFAIK there is only one real problem left that will keep many big businesses from deploying IPv6: multi-homing. The technology to have more than one upstream provider for IPv6 connectivity is still in flux. /32 allocated which they can advertise on as many providers as they like.
AFAICT for big buisnesses this is a non-issue, they will already be an "autonomous system" and will probablly have no trouble getting a globally routable
For small and medium buisnesses there is a big problem though. With IPV4 you could use NAT and I belive you could also advertise space allocated by one provider on another provider. With ipv6 NAT is not readilly availible (linux doesn't support it for philosophical reasons and I don't think many other platforms support it either) and I don't think you are allowed to advertise space from one provider on another provider.
There was a system called A6 which was supposed to simplify handling of sites with multiple addresses running in paralell but it was extremely complicated in itself and has been abandoned afaict.
Just browsing through quickly, it seems that at least the extreme models ship with 4gb of RAM + video card memory. Assuming they use 32 bit window - due to the better drivers, from what I've seen it benchmarks better than 64 bit, isn't this completly pointless?
under an OS that only supports 4GB of address space 4GB of ram tends to give you a little extra usable memory over 3GB (how much depends on the other hardware in the box). Also with some chipsets using a matched pair gives better performance than an unmatched pair.
It's probablly a marginal benefit but given the extremely low cost of ram now some may consider it worth having.
I guess it depends on what you consider a hardware upgrade. Huge numbers of games for older consoles had banking hardware and saveram (some of which may be used for things other than saves) in the cart and occasionally there was other stuff in there too.
DS carts on the other hand have an interface that is disk like rather than rom like so if you want extra ram you need to put it on the GBA slot. That means an extra cart for the user to lose. So far the only commercial DS product i'm aware of that does so is the web browser.
Using escaping to protect your queries is like doing strings by manually allocating blocks of memory and then storing a sequence of characters in them followed by a null terminator (either directly or through a series of helper functions).
Both can sometimes be forced on you by the environment you are working in. Both require an extreme level of attention to detail by both the initial programmer and later programmers who work on the code. Both can easilly lead to security holes if the programmer makes a simple mistake.
Houses have pretty big safety margins, if most of the house is built to code the odd bad beam or joint isn't going to bring things crashing down.
A driver is a peice of software written in a language without safety checks, running in an environment without process seperation, having to deal with obscure concepts like IRQL and often interacting directly with hardware. One bug can easilly bring the system crashing down and the chance of a test team finding every bug is pretty low even if they have access to read the source.
there's absolutely nothing performance critical enough to justify running it in kernel space.
no but moving to a usermode driver would basically mean a complete rewrite and would reduce the number of windows versions supported. Doesn't seem very attractive to me.
The drivers are USB protocol drivers -- they run in user space.
They can be written as userspace drivers but they probablly aren't. Moving your driver to userspace means rewriting it and losing compatibility with a number of older windows versions. Not a very attractive proposition IMO.
USB drivers live in userspace (only the generic read/write support for USB devices live in kernel space)
Afaict on windows most USB drivers live in kernel space. It is possible to make user mode USB drivers but if you do afaict they will not work on 2K or server 2K3 (this may have changed but if it has MS didn't bother to update thier faq and I don't have access to download the wdk anymore). Furthermore a user mode driver can't have kernel mode clients (which means you can't do keyboard/mouse drivers or mass storage drivers in userspace)