I see two problems with this kind of approach though
1: the code may get triggered by accident leading to a legitimate user getting frustrated at the games apparent buginess/uncompletability. 2: pirates may not realise that the problems they are experiencing are a result of antipiracy meausres.
Either way you have users who think the game is buggy as hell telling their friends to avoid it.
But isn't PayPal based in Luxembourg? "PayPal (Europe) S.à r.l. & Cie, S.C.A" is based there. "Paypal inc" (which afaict is both the parent paypal company and the US paypal operation) is based in California.
Only at one place in my post did I assume TCP/IP (a reasonable assumption IMO based on current trends). That was the need for a configuration mecahnism because not all networks had DHCP. All the other issues I listed apply regardless of what protocol is used over ethernet.
As long as they give a public IP to anyone who bitches too much and/or offer a public IP at a small (initially) extra charge I think a lawsuit is unlikely.
And on what grounds would such a suit be brought anyway? is there any legal definition of internet connection? do the contracts promise a public IP?
I don't see anything in this thread talking about a situation were all connections are natted. Such a situation is clearly unfeasible.
Natting home lusers (I suspect geeks will have a choice of either putting up with it too of paying extra) should be sufficiant for a while at least. Proper servers will almost certainly continue to get public IPs.
Note that how well this sort of thing (you have the details of the method off a little but that isn't really relavent to the discussion) works depends on the type of nat.
With a full cone, restricted cone or restricted port cone NAT the technique will work provided it is properly implemented and both sides start sending beacon packets before expecting anything from the other side..
With a "symetric NAT" that treats every local IP/local port/remote IP/remote port combination as a "connnection" and uses unrelated ports on the WAN side for each connection (even if the source IP/PORT is the same) it won't work at all since the matchmaking service has no way of finding out what natted local port will be the source of the packets sent to the other peer.
If the nat is a "port preservative" type (a type that tries to make the natted local port the same as the un-natted local port but changes the local port to avoid conflicts) the technique will work some but not all of the time and the chance of failure will increase with the load on the NAT. If the NAT is round robining public IPs then depending on how that is handled that could also cause problems.
In summary the technique works most of the time with consumer NAT boxes that are generally under light load but things could be much worse with "carrier grade NAT" setups handling large numbers of users. Particually if they are stingy on the ratio of public IPs to private IPs.
So, I keep hearing all this news about them running low... What happens when we run out? My bet is that ISPs force end lusers behind ISP level NAT to free up the IPs for more profitable customers.
But not significantly more, all the important stuff is available on V4 and will be for the forseeable future. Likewise I would expect most users to have access to those resources through some mechanism (likely some form of NAT).
And your V6 connection is likely to be slower than your V4 one at communicating with a given resource, especially if your V6 connection comes in the form of a third party tunnel.
So other than P2P (which the ISPs hate anyway) there is little incentive to implement V6.
My house has a single phase, 100kW maximum supply, this is pretty normal in the UK. No it isn't, I think you are confusing amps with kilowatts. Typical in the UK is about 60A-100A single phase which at 240V works out to 14-24KW
100KW would be about 400A single phase or 138A three phase.
The article implies though it doesn't explicitly state that the package was sent via a normal fedex service rtather than one of thier special ones. There isn't much a carrier can do if an idiot sends something by an unsutiable service and fails to make them aware of the packages true contents until after the package is lost!
Umm, Apple have used VGA and then DVI and then Mini DisplayPort for video interfaces in the past 10 years. They had mini versions of these connectors, for which you were also given the necessary dongle to upsize it to the standard version of the interface. No you aren't, it's bundled with the mini but on the laptops i'm pretty sure it's an optional extra (at least it was when I bought mine)
But even if you spring for the adaptor it's still another thing to lose/get annoyed by and the only place you are going to pick one up in a B&M outlet is probablly an apple store.
And apple laptops have been all over the place with display connectors. IIRC the ibook used mini-vga (Not sure what the powerbooks did), the first macbooks used mini DVI, the first macbook pros used full sized dvi, the first macbook air used micro DVI and then they moved to them all to mini displayport with the unibody stuff. Meanwhile the de-facto standard for lecture theatre/confrence room projector setups has remained VGA so every apple laptop needed an adaptor and there is a good chance it would be fifferent from the last one.
And worse still the adaptors to DVI are DVI-D only so you can't combine them with the DVI to VGA adaptor you have hanging arround you have to buy a seperate adaptor from whatever apple mini plug you have to VGA (and good f*cking luck buying such an adaptor anywhere except an online store or an official apple reseller).
And then there is the clusterfuck that was the mini displayport to dual link DVI adaptor (read the reviews on it...) or the clusterfuck of being unable to buy mini displayport to dislayport adaptors (they do exist now but the only seller i've found is monoprice in the US).
What's the incremental cost on adding a network card? $0.10? $0.05? Depends what is there already. The ethernet chip, magnetics and sockets probablly aren't hugely expensive in quantity given that whole cards sell for $10 or so. I'd still think we are talking dollars rather than cents though.
Though IMO the real problems with ethernet as an interface aren't the cost of the interface but the following 1: At least in an office environment it's actually useful to have a demarcation between "trusted peripherals" and the "big wide network". Would you want all the gear on your desk accessible to everyone in your building or more. Yes you CAN have multiple seperate networks (and I do because I use ethernet based test gear) but that's not a solution I'd recommend with end lusers and in any case most machines only have one network port. 2: ethernet perhipherals generally need far more intelligence. A usb hard drive just presents itself as a block device and lets the OS handle everything else. An ethernet one has to security (because it might not be on a trusted network), filesystems (because people expect ethernet devices to be multiuser capable) etc. That means instead of a cheap USB-sata bridge chip you are talking a complete computer in there. 3: ethernet perhipherals need a means of configuration since the network they are on may not offer DHCP and most ethernet perhipherals will also want security options. While there are ways to do said intital configuration over the network it's often painful (for example connecting it directly to a PC and messing with the network settings on said PC so that you can talk to the thing and set the settings it needs to operate on your network).
Now for some gear ethernet works out well. A big laser printer for example will already have a fairly powerful CPU. It will already have a front panel interface and in a small office situation people will probablly want it accessible to everyone in the office anyway. In a larger business IT should (though unfortunately should doesn't always mean will:/ ) be responsible for appropriately securing access to it.
P.S. versions of ethernet currently in desktop use are not fast enough to deliver a new frame to a display 60 times per second
I've been using something similar at work for a long, long time: A USB 2.0 docking station.
Included are PS/2 mouse and keyboard ports, a USB hub with two ports, RS-232, parallel, and 10/100 Ethernet. Yeah if you don't care too much about network performance (or don't have gigabit anyway) and you don't use and external monitor (or don't care about the low performance that comes from USB based video) then USB will do it.
USB 3 should improve matters but it will still be cuttting things a bit close to the line for high performance video especially as on many systems USB 3 is stuck behind PCIe 1.0 x1 ( 1920×1200×24×60=3317760000 bits per second vs a "raw speed" of 4Gbps for USB 3 and 2.5Gbps for PCIe 1.0 x1),. I don't think i've actually seen anywhere selling a USB 3 video adaptor yet nor do most laptops seem to offer it.
Expresscard docking stations also exist and should offer better video performance than USB but they mean a bulky card plugging into the laptop with a thick cable and again the bandwidth limitations of PCIe 1.0 x1 will rear their ugly head if you actually try to push 60FPS fullscreen video on them.
I understand the former may have had to do with previous hardware versions that could not tolerate operating with the cover closed, but that's a pretty lousy excuse. It's really a result of apple's obsession with smooth sleek stuff. The macbooks have no vent holes on the bottom, this has it's good points (looks sleek when closes and no need to worry about blocking them by using the machine on a poor surface) but it means the only real place for the machine to vent through is the keyboard and hence running it closed (i've done it under linux) is a bad idea.
Plus afaict very few machines actually had a "USB chipset", intel (and I think AMD too but I haven't followed that side of the fence closely) bundled USB in their southbridges whether you wanted it or not.
Indeed on the minis they are more generous with adaptors than on the laptops and imacs (I dunno what the situation is the with pro). Presumablly because you need to hook up a monitor to use the thing at all.
Plus I don't see unifying the network and local interconnects as desirable anyway at least in a corporate situation. The local interconnect is "trusted" (anyone who has access to it probablly has physical access to the PC anyway) and is autoconfigured. The network spans a whole site or more, is relatively untrusted (since anyone anywhere on the site can connect something to it) and is managed by the IT team.
In a server or other technical situation (e.g. lab test equipment) one often has multiple seperate networks running the same protocols/hardware. but doing that on the lusers desks is just asking for stuff to get connected to the wrong network.
I can, people do unapproved things all the time either because they are ignorant of the potential consequences or because they thing (rightly or wrongly) that the probability is low enough that they can get away with it.
Look at Grand Theft Auto 3 or Vice City. You discover that other than special objects, like police chasing you or cars for missions, nothing exists that is out of your FOV. Look at a street, turn around, then turn back. It changes because it is regenerated. Again, done because of the small memory on the PS2 which it also runs on. OTOH without some form of regneeration in a game like that (not sure what the 2D GTAs were like) the entire city would grind to a halt from the mayhem the player left behind.
True to an extent though i've noticed that a LOT less games are offering single console multiplayer this generation now that consoles have online support.
This increased initial window is optional: a TCP MAY start with a larger initial window. However, we expect that most general-purpose TCP implementations would choose to use the larger initial congestion window given in equation (1) above.
new versions of windows come out every 7-8 years While your point in general is correct you are exaggerating. Looking at the years of windows releases (I could look at the months but I CBA to and it dosen't change the overall point) and ignoring server releases.
conventional series 1.0: 1985 2.0: 1988: 3 years from previous release 3.0: 1990: 2 years from previous release 3.1: 1992: 2 years from previous release 95: 1995: 3 years from previous release 98: 1998: 3 years from previous release ME: 2000: 2 years from previous release NT series 3.1: 1993 3.5: 1994 1 year from previous release 3.51: 1995 1 year from previous release 4.0: 1996 1 year from previous release 2000: 2000 4 years from previous release merged series XP: 2001 1 year from previous release Vista: 2007: 6 years from previous release 7: 2009: 2 years from previous release
Looks to me like mostly 1-3 years with a few outliers.
use this in practice on a debian (stable) web server, with a few select web apps such as wordpress pegged to unstable. For those reading along be aware that while this may be workable for webapps (which are usually written in scripting languages) it can be a poor strategy for packages in general because often apps in unstable often pick up dependencies on unstable's versions of key libraries (this isn't as bad as it used to be due to the introduction of symbols files but it's still an issue)
use this in practice on a debian (stable) web server, with a few select web apps such as wordpress pegged to unstable. For those reading along be aware that while this may be workable for webapps (which are usually written in scripting languages) it can be a poor strategy for packages in general because often apps in unstable often pick up dependencies on unstable's versions of key libraries (this isn't as bad as it used to be due to the introduction of symbols files but it's still an issue).
Slashdot Mirror
I see two problems with this kind of approach though
1: the code may get triggered by accident leading to a legitimate user getting frustrated at the games apparent buginess/uncompletability.
2: pirates may not realise that the problems they are experiencing are a result of antipiracy meausres.
Either way you have users who think the game is buggy as hell telling their friends to avoid it.
But isn't PayPal based in Luxembourg?
"PayPal (Europe) S.à r.l. & Cie, S.C.A" is based there. "Paypal inc" (which afaict is both the parent paypal company and the US paypal operation) is based in California.
Only at one place in my post did I assume TCP/IP (a reasonable assumption IMO based on current trends). That was the need for a configuration mecahnism because not all networks had DHCP. All the other issues I listed apply regardless of what protocol is used over ethernet.
As long as they give a public IP to anyone who bitches too much and/or offer a public IP at a small (initially) extra charge I think a lawsuit is unlikely.
And on what grounds would such a suit be brought anyway? is there any legal definition of internet connection? do the contracts promise a public IP?
I don't see anything in this thread talking about a situation were all connections are natted. Such a situation is clearly unfeasible.
Natting home lusers (I suspect geeks will have a choice of either putting up with it too of paying extra) should be sufficiant for a while at least. Proper servers will almost certainly continue to get public IPs.
Note that how well this sort of thing (you have the details of the method off a little but that isn't really relavent to the discussion) works depends on the type of nat.
With a full cone, restricted cone or restricted port cone NAT the technique will work provided it is properly implemented and both sides start sending beacon packets before expecting anything from the other side..
With a "symetric NAT" that treats every local IP/local port/remote IP/remote port combination as a "connnection" and uses unrelated ports on the WAN side for each connection (even if the source IP/PORT is the same) it won't work at all since the matchmaking service has no way of finding out what natted local port will be the source of the packets sent to the other peer.
If the nat is a "port preservative" type (a type that tries to make the natted local port the same as the un-natted local port but changes the local port to avoid conflicts) the technique will work some but not all of the time and the chance of failure will increase with the load on the NAT. If the NAT is round robining public IPs then depending on how that is handled that could also cause problems.
In summary the technique works most of the time with consumer NAT boxes that are generally under light load but things could be much worse with "carrier grade NAT" setups handling large numbers of users. Particually if they are stingy on the ratio of public IPs to private IPs.
So, I keep hearing all this news about them running low... What happens when we run out?
My bet is that ISPs force end lusers behind ISP level NAT to free up the IPs for more profitable customers.
But not significantly more, all the important stuff is available on V4 and will be for the forseeable future. Likewise I would expect most users to have access to those resources through some mechanism (likely some form of NAT).
And your V6 connection is likely to be slower than your V4 one at communicating with a given resource, especially if your V6 connection comes in the form of a third party tunnel.
So other than P2P (which the ISPs hate anyway) there is little incentive to implement V6.
My house has a single phase, 100kW maximum supply, this is pretty normal in the UK.
No it isn't, I think you are confusing amps with kilowatts. Typical in the UK is about 60A-100A single phase which at 240V works out to 14-24KW
100KW would be about 400A single phase or 138A three phase.
The article implies though it doesn't explicitly state that the package was sent via a normal fedex service rtather than one of thier special ones. There isn't much a carrier can do if an idiot sends something by an unsutiable service and fails to make them aware of the packages true contents until after the package is lost!
Afaict while the standard haven't officially been ratified for running eSATA at 6Gbps yet there is nothing to actually prevent it.
Also IIRC on many boards the USB 3 is stuck behind PCIe 1.0 x1.
Umm, Apple have used VGA and then DVI and then Mini DisplayPort for video interfaces in the past 10 years. They had mini versions of these connectors, for which you were also given the necessary dongle to upsize it to the standard version of the interface.
No you aren't, it's bundled with the mini but on the laptops i'm pretty sure it's an optional extra (at least it was when I bought mine)
But even if you spring for the adaptor it's still another thing to lose/get annoyed by and the only place you are going to pick one up in a B&M outlet is probablly an apple store.
And apple laptops have been all over the place with display connectors. IIRC the ibook used mini-vga (Not sure what the powerbooks did), the first macbooks used mini DVI, the first macbook pros used full sized dvi, the first macbook air used micro DVI and then they moved to them all to mini displayport with the unibody stuff. Meanwhile the de-facto standard for lecture theatre/confrence room projector setups has remained VGA so every apple laptop needed an adaptor and there is a good chance it would be fifferent from the last one.
And worse still the adaptors to DVI are DVI-D only so you can't combine them with the DVI to VGA adaptor you have hanging arround you have to buy a seperate adaptor from whatever apple mini plug you have to VGA (and good f*cking luck buying such an adaptor anywhere except an online store or an official apple reseller).
And then there is the clusterfuck that was the mini displayport to dual link DVI adaptor (read the reviews on it...) or the clusterfuck of being unable to buy mini displayport to dislayport adaptors (they do exist now but the only seller i've found is monoprice in the US).
What's the incremental cost on adding a network card? $0.10? $0.05?
Depends what is there already. The ethernet chip, magnetics and sockets probablly aren't hugely expensive in quantity given that whole cards sell for $10 or so. I'd still think we are talking dollars rather than cents though.
Though IMO the real problems with ethernet as an interface aren't the cost of the interface but the following
1: At least in an office environment it's actually useful to have a demarcation between "trusted peripherals" and the "big wide network". Would you want all the gear on your desk accessible to everyone in your building or more. Yes you CAN have multiple seperate networks (and I do because I use ethernet based test gear) but that's not a solution I'd recommend with end lusers and in any case most machines only have one network port.
2: ethernet perhipherals generally need far more intelligence. A usb hard drive just presents itself as a block device and lets the OS handle everything else. An ethernet one has to security (because it might not be on a trusted network), filesystems (because people expect ethernet devices to be multiuser capable) etc. That means instead of a cheap USB-sata bridge chip you are talking a complete computer in there.
3: ethernet perhipherals need a means of configuration since the network they are on may not offer DHCP and most ethernet perhipherals will also want security options. While there are ways to do said intital configuration over the network it's often painful (for example connecting it directly to a PC and messing with the network settings on said PC so that you can talk to the thing and set the settings it needs to operate on your network).
Now for some gear ethernet works out well. A big laser printer for example will already have a fairly powerful CPU. It will already have a front panel interface and in a small office situation people will probablly want it accessible to everyone in the office anyway. In a larger business IT should (though unfortunately should doesn't always mean will :/ ) be responsible for appropriately securing access to it.
P.S. versions of ethernet currently in desktop use are not fast enough to deliver a new frame to a display 60 times per second
I've been using something similar at work for a long, long time: A USB 2.0 docking station.
Included are PS/2 mouse and keyboard ports, a USB hub with two ports, RS-232, parallel, and 10/100 Ethernet.
Yeah if you don't care too much about network performance (or don't have gigabit anyway) and you don't use and external monitor (or don't care about the low performance that comes from USB based video) then USB will do it.
USB 3 should improve matters but it will still be cuttting things a bit close to the line for high performance video especially as on many systems USB 3 is stuck behind PCIe 1.0 x1 ( 1920×1200×24×60=3317760000 bits per second vs a "raw speed" of 4Gbps for USB 3 and 2.5Gbps for PCIe 1.0 x1) ,. I don't think i've actually seen anywhere selling a USB 3 video adaptor yet nor do most laptops seem to offer it.
Expresscard docking stations also exist and should offer better video performance than USB but they mean a bulky card plugging into the laptop with a thick cable and again the bandwidth limitations of PCIe 1.0 x1 will rear their ugly head if you actually try to push 60FPS fullscreen video on them.
I understand the former may have had to do with previous hardware versions that could not tolerate operating with the cover closed, but that's a pretty lousy excuse.
It's really a result of apple's obsession with smooth sleek stuff. The macbooks have no vent holes on the bottom, this has it's good points (looks sleek when closes and no need to worry about blocking them by using the machine on a poor surface) but it means the only real place for the machine to vent through is the keyboard and hence running it closed (i've done it under linux) is a bad idea.
Plus afaict very few machines actually had a "USB chipset", intel (and I think AMD too but I haven't followed that side of the fence closely) bundled USB in their southbridges whether you wanted it or not.
Indeed on the minis they are more generous with adaptors than on the laptops and imacs (I dunno what the situation is the with pro). Presumablly because you need to hook up a monitor to use the thing at all.
Plus I don't see unifying the network and local interconnects as desirable anyway at least in a corporate situation. The local interconnect is "trusted" (anyone who has access to it probablly has physical access to the PC anyway) and is autoconfigured. The network spans a whole site or more, is relatively untrusted (since anyone anywhere on the site can connect something to it) and is managed by the IT team.
In a server or other technical situation (e.g. lab test equipment) one often has multiple seperate networks running the same protocols/hardware. but doing that on the lusers desks is just asking for stuff to get connected to the wrong network.
I can, people do unapproved things all the time either because they are ignorant of the potential consequences or because they thing (rightly or wrongly) that the probability is low enough that they can get away with it.
Look at Grand Theft Auto 3 or Vice City. You discover that other than special objects, like police chasing you or cars for missions, nothing exists that is out of your FOV. Look at a street, turn around, then turn back. It changes because it is regenerated. Again, done because of the small memory on the PS2 which it also runs on.
OTOH without some form of regneeration in a game like that (not sure what the 2D GTAs were like) the entire city would grind to a halt from the mayhem the player left behind.
True to an extent though i've noticed that a LOT less games are offering single console multiplayer this generation now that consoles have online support.
This increased initial window is optional: a TCP MAY start with a larger initial window. However, we expect that most general-purpose TCP implementations would choose to use the larger initial congestion window given in equation (1) above.
new versions of windows come out every 7-8 years
While your point in general is correct you are exaggerating. Looking at the years of windows releases (I could look at the months but I CBA to and it dosen't change the overall point) and ignoring server releases.
conventional series
1.0: 1985
2.0: 1988: 3 years from previous release
3.0: 1990: 2 years from previous release
3.1: 1992: 2 years from previous release
95: 1995: 3 years from previous release
98: 1998: 3 years from previous release
ME: 2000: 2 years from previous release
NT series
3.1: 1993
3.5: 1994 1 year from previous release
3.51: 1995 1 year from previous release
4.0: 1996 1 year from previous release
2000: 2000 4 years from previous release
merged series
XP: 2001 1 year from previous release
Vista: 2007: 6 years from previous release
7: 2009: 2 years from previous release
Looks to me like mostly 1-3 years with a few outliers.
use this in practice on a debian (stable) web server, with a few select web apps such as wordpress pegged to unstable.
For those reading along be aware that while this may be workable for webapps (which are usually written in scripting languages) it can be a poor strategy for packages in general because often apps in unstable often pick up dependencies on unstable's versions of key libraries (this isn't as bad as it used to be due to the introduction of symbols files but it's still an issue)
use this in practice on a debian (stable) web server, with a few select web apps such as wordpress pegged to unstable.
For those reading along be aware that while this may be workable for webapps (which are usually written in scripting languages) it can be a poor strategy for packages in general because often apps in unstable often pick up dependencies on unstable's versions of key libraries (this isn't as bad as it used to be due to the introduction of symbols files but it's still an issue).