1) I'm skeptical this will force MS into anything in regard to the GPLv3, except/maybe/ if MS continues to distribute coupons after something critical goes GPLv3, in part because I think it likely a court will hold that the agreement was made before the license change and MS can't be held to an ex pos facto license change, and in part because it's really still up in the air whether what they are doing would legally be distribution or not -- it may not be, and no opinions posted here are likely to matter when it comes time for a court to rule on that.
2) With draft-3, Linus has said he's actually quite pleased with how it turned out, now. It remains to be seen whether he'll like the legally binding final version or not, but it's looking rather better than it did. Of course, he's not going to change the license all on his own, either, but his opinion will make a big difference, and even if it didn't, if his opinion on it has changed that much, others may have as well.
3) It'd be a long process, but the kernel/could/ change to GPLv3, regardless of what many say,/if/ most/all of the current major contributors were to agree to it. It'd be quite a long journey, several years, but if all the major contributors agreed that's a lot of code that could be switched first to GPLv2 or 3, and later to v3 only, without any rewriting at all. Given the existing rate of code turnover, within three years or so I'd say a good 2/3 to 3/4 of the code could be GPLv2/GPLv3 dual licensed with little effort at all. After another couple years of more concentrated effort, culminating in a seriously focused coding marathon much like the Bitkeeper>GIT transition (which many thought couldn't be done either, at least and keep the same productivity), the remaining code should have been rewritten, and the GPLv2 side of the dual license could be dropped.
The license couldn't be turned on a dime, and it WOULD take the agreement of nearly all current and former major contributors, but it COULD be changed, given a time frame of several years and sufficient motivation to do so.
4) I've not yet read the Groklaw article (I will), but the point you quote about the patent protection only applying if the kernel is GPLv3 is valid -- for the kernel, not for Novel/SuSE. IOW, for the patent protection to apply to/the//kernel/ the kernel will have to be GPLv3 licensed. The protection in the license can only apply to something licensed under it, so of course, it could only apply to the kernel if it's so licensed. However, as you point out, that could be a VERY good reason to switch to GPLv3, quite apart from any other reason. The way patents are going, it may be that the kernel will ultimately/have/ to switch, or lose the momentum it currently has to something else (say Sun's Solaris kernel, if they license it GPLv3 as they are making noises saying they might do). It could well be that within a few years, few businesses will risk deployment of a GPLv2 kernel due to perceived (note that word, perceived, they may or may not have been legally validated) patent issues, and the money now funding all those full time Linux kernel devs will be funding something else (Solaris development?) if the Linux kernel isn't in the process of going GPLv3 by then.
5) During the dual GPLv2/GPLv3 license period, and this applies to anything licensed GPLv2 or later as well, the license says at YOUR option, meaning each user/distributor can choose the version they wish. Some have claimed that such would allow users to force distributors to the v3 terms if they wished. That isn't so, and the FSF doesn't hold it to be so either. It simply means that anyone downstream can redistribute with only one of the licenses if so desired, rather than both. It does NOT mean anyone can hold a specific distributor to the terms of one or the other licenses if they choose to be held to the other one. Thus, MS (if it is held to
Ouch! I knew there was a reason I was uneasy about those integrated modem/router units! I hadn't gotten one mainly because I didn't want to have to replace both when I replaced one, but I was a bit worried about the flexibility in general as well. Now I see there was a reason to be. Unfortunately, that's not of much help to you. Hopefully, you won't need need to worry about it again for awhile, however.
Well, maybe the info will be of help to others. I know I've followed from google and found of help a few/. threads over the years.
I'm on cable as well (Cox), and the account is associated with the MAC address of the modem,/not/ the MAC address of the computer, which gets its own IP address.
With some simplification, it works this way. (I talk about Ethernet below, but the USB interface is similar, within reason, of course.) They lock the account to the modem's MAC address. When the modem connects, after physically syncing, it asks for an IP address, of course giving the MAC address in the lower layer connection. If the modem's MAC address is registered, the assigned IP address (for the modem only) is a non-publicly-routable IP address (Cox uses addresses in the 10./8 range, but I'm not sure if that's DOCSIS specified or not, so it may differ elsewhere). That gets the modem online, but you can't yet connect, because your computer or router doesn't have an IP yet.
The modem, still setting up, then requests such things as the TOD (time of day) and a config file. The config file will contain the settings for your account, including your allowed bandwidth caps, which the modem then enforces (this is why the cablecos are so insistent on controlling firmware and the like, it can normally be flashed only from the cable side, because it's the modem that does the actual capping), and certain modem-side filters (port 80 inbound, among others, on Cox, for residential accounts). One particular config file setting of interest is the number of customer-side MAC addresses it can allow thru. Often, this is only one, but it may be more, if the account policy is set for more (many cablecos sell access for additional computers, really, access for more than one MAC address, for an additional fee).
If you have a Motorola Surfboard series modem (other brands have the interface, but lack the level of info on it that the Motorolas provide), surf over to http://192.168.100.1/ (which you may recognize is another non-publicly routable address), and on one of the pages, you'll see how many MAC addresses the modem has learned and how many you are allowed.
Then, after the modem is configured, it turns on the bridging mode between the Ethernet and RFC side and lets whatever is connected thru to the cableco and Internet -- but ONLY up to the number of allowed MAC addresses. It will "learn" the first MAC addresses it sees, ignoring any others on the LAN.
The caveat in switching MAC addresses should be apparent -- if you've already reached your configured limit (normally a single MAC address), in ordered to get your new MAC address to connect, you'll have to reboot the modem, because it learned the other MAC address and until that memory is erased, it won't allow any more thru to get IPs from the cableco.
Once your router or computer is allowed to connect, it will ask for and get an IP address from the cableco, normally the same one it got the last time it asked for an address, based on the MAC making the request. If the MAC is different, you therefore get a different IP address.
Thus, to force an IP address change, you change the MAC address you are using to connect, then reboot the modem so it forgets the old one and allows the new one thru. You should then get a new IP address. Eventually, the allotted IP addresses available for assignment will be used up, and the DHCP server will start reusing ones that aren't in use. However, that typically takes quite some time, because there's a significant number of "free" IPs, and because few people cycle MAC addresses. Thus, the system is likely to remember the MAC off your old Ethernet card you used a year ago, even tho it hasn't been connected since, and still give you the same IP address, unless of course you artificially set a different MAC address.
That brings us back to where we were before, the problem of setting a different MAC address. It's easier on Linux (and I believe the various BSDs) than on MSWormOS, but it should be possible in any case, even if you have to get a new NIC to do it. A router should make
Frist psot with my new account =8^), tho I've lurked and occasionally posted AC for years.
Your ISP is likely associating your assigned IP with your MAC address, as is quite common with DHCP, giving users that want a stable address at least a bit of stability. This seems to be particularly common with cable modem systems. Change that, and you get a new IP address. It's likely that easy. =8^)
MAC addresses are normally hard-burned into the hardware (Ethernet and the like), and are supposed to function as a GUID (globally unique ID), so changing your NIC is one way to change your MAC address. However, there's often a way to set them to something other than the hardware MAC address manually. The idea is that a network may have configured a specific association (like IP address, or even permission to connect, some ISPs register MAC addresses and won't allow unregistered addresses a connection or give them an IP address, MAC address filtering is also very common on wireless networks, for security reasons) that may be inconvenient to change when you switch NICs or the like, and the ability to manually set a MAC address allows one to set the same one they previously used.
How you change your MAC address, however, will obviously depend on what OS and/or Ethernet/other-connectivity drivers you use. Here on Gentoo, there's a network services module that combined with an app called macchanger, allows me to set my MAC at will, every time I bring up the connection on that interface. I have it set to entirely random, so if my ISP is tracking it, I might appear to be connecting with a multigigabit Internet backbone router one day, and an old 2 Mbit thinnet card or whatever the next, but that could cause problems if the ISP was relying on that for something, in which case I could just set it to randomize within my hardware type or just the specific NIC manufacturer. Anyway, with my randomized MAC address, I get a different IP address every time I reconnect (tho the computer keeps the same MAC and gets the same IP when I simply hibernate, aka suspend to disk). Otherwise, as yours, my ISP continues to hand people the same MAC associated IP for sometimes/years/.
On MSWormOS, I believe implementation is left to the Ethernet driver. If your NIC driver implements manually setting your MAC, it should have such a setting in the appropriate properties tab. Else you may be able to set it in the registry, if you know what you are doing. I stick to freedomware (see the sig) these days, so can't give you much more on that.
Routers often have an option to clone the MAC of a connected computer on the LAN to the WAN side, again, in case the ISP specifically authenticates in part by registered MAC address and won't allow changing it without calling them. Others let you set it specifically. Some, particularly those running OpenWRT or similar firmware, may allow fully randomized MAC addresses, much as I described I do with my Gentoo system above. You mentioned that switching between direct connect and using your router didn't change your assigned IP address, but it's possible that's because the router was already cloning your computer's MAC address to the WAN side, so the ISP saw no change.
So anyway, try changing your MAC address. You'll likely get a different IP address that way. If you don't, the worst that should happen is they won't let you connect, and you change it back. Of course, if they associate the assigned address with the login info (PPPoE or the like) or MAC address of the modem, you'll still connect, but changing the MAC address on your computer or router won't cause the IP address they assign you to change.
Slashdot Mirror
Five points.
/maybe/ if MS continues to distribute coupons after something critical goes GPLv3, in part because I think it likely a court will hold that the agreement was made before the license change and MS can't be held to an ex pos facto license change, and in part because it's really still up in the air whether what they are doing would legally be distribution or not -- it may not be, and no opinions posted here are likely to matter when it comes time for a court to rule on that.
/could/ change to GPLv3, regardless of what many say, /if/ most/all of the current major contributors were to agree to it. It'd be quite a long journey, several years, but if all the major contributors agreed that's a lot of code that could be switched first to GPLv2 or 3, and later to v3 only, without any rewriting at all. Given the existing rate of code turnover, within three years or so I'd say a good 2/3 to 3/4 of the code could be GPLv2/GPLv3 dual licensed with little effort at all. After another couple years of more concentrated effort, culminating in a seriously focused coding marathon much like the Bitkeeper>GIT transition (which many thought couldn't be done either, at least and keep the same productivity), the remaining code should have been rewritten, and the GPLv2 side of the dual license could be dropped.
/the/ /kernel/ the kernel will have to be GPLv3 licensed. The protection in the license can only apply to something licensed under it, so of course, it could only apply to the kernel if it's so licensed. However, as you point out, that could be a VERY good reason to switch to GPLv3, quite apart from any other reason. The way patents are going, it may be that the kernel will ultimately /have/ to switch, or lose the momentum it currently has to something else (say Sun's Solaris kernel, if they license it GPLv3 as they are making noises saying they might do). It could well be that within a few years, few businesses will risk deployment of a GPLv2 kernel due to perceived (note that word, perceived, they may or may not have been legally validated) patent issues, and the money now funding all those full time Linux kernel devs will be funding something else (Solaris development?) if the Linux kernel isn't in the process of going GPLv3 by then.
1) I'm skeptical this will force MS into anything in regard to the GPLv3, except
2) With draft-3, Linus has said he's actually quite pleased with how it turned out, now. It remains to be seen whether he'll like the legally binding final version or not, but it's looking rather better than it did. Of course, he's not going to change the license all on his own, either, but his opinion will make a big difference, and even if it didn't, if his opinion on it has changed that much, others may have as well.
3) It'd be a long process, but the kernel
The license couldn't be turned on a dime, and it WOULD take the agreement of nearly all current and former major contributors, but it COULD be changed, given a time frame of several years and sufficient motivation to do so.
4) I've not yet read the Groklaw article (I will), but the point you quote about the patent protection only applying if the kernel is GPLv3 is valid -- for the kernel, not for Novel/SuSE. IOW, for the patent protection to apply to
5) During the dual GPLv2/GPLv3 license period, and this applies to anything licensed GPLv2 or later as well, the license says at YOUR option, meaning each user/distributor can choose the version they wish. Some have claimed that such would allow users to force distributors to the v3 terms if they wished. That isn't so, and the FSF doesn't hold it to be so either. It simply means that anyone downstream can redistribute with only one of the licenses if so desired, rather than both. It does NOT mean anyone can hold a specific distributor to the terms of one or the other licenses if they choose to be held to the other one. Thus, MS (if it is held to
Ouch! I knew there was a reason I was uneasy about those integrated modem/router units! I hadn't gotten one mainly because I didn't want to have to replace both when I replaced one, but I was a bit worried about the flexibility in general as well. Now I see there was a reason to be. Unfortunately, that's not of much help to you. Hopefully, you won't need need to worry about it again for awhile, however.
/. threads over the years.
Well, maybe the info will be of help to others. I know I've followed from google and found of help a few
Thanks,
Duncan
I'm on cable as well (Cox), and the account is associated with the MAC address of the modem, /not/ the MAC address of the computer, which gets its own IP address.
With some simplification, it works this way. (I talk about Ethernet below, but the USB interface is similar, within reason, of course.) They lock the account to the modem's MAC address. When the modem connects, after physically syncing, it asks for an IP address, of course giving the MAC address in the lower layer connection. If the modem's MAC address is registered, the assigned IP address (for the modem only) is a non-publicly-routable IP address (Cox uses addresses in the 10./8 range, but I'm not sure if that's DOCSIS specified or not, so it may differ elsewhere). That gets the modem online, but you can't yet connect, because your computer or router doesn't have an IP yet.
The modem, still setting up, then requests such things as the TOD (time of day) and a config file. The config file will contain the settings for your account, including your allowed bandwidth caps, which the modem then enforces (this is why the cablecos are so insistent on controlling firmware and the like, it can normally be flashed only from the cable side, because it's the modem that does the actual capping), and certain modem-side filters (port 80 inbound, among others, on Cox, for residential accounts). One particular config file setting of interest is the number of customer-side MAC addresses it can allow thru. Often, this is only one, but it may be more, if the account policy is set for more (many cablecos sell access for additional computers, really, access for more than one MAC address, for an additional fee).
If you have a Motorola Surfboard series modem (other brands have the interface, but lack the level of info on it that the Motorolas provide), surf over to http://192.168.100.1/ (which you may recognize is another non-publicly routable address), and on one of the pages, you'll see how many MAC addresses the modem has learned and how many you are allowed.
Then, after the modem is configured, it turns on the bridging mode between the Ethernet and RFC side and lets whatever is connected thru to the cableco and Internet -- but ONLY up to the number of allowed MAC addresses. It will "learn" the first MAC addresses it sees, ignoring any others on the LAN.
The caveat in switching MAC addresses should be apparent -- if you've already reached your configured limit (normally a single MAC address), in ordered to get your new MAC address to connect, you'll have to reboot the modem, because it learned the other MAC address and until that memory is erased, it won't allow any more thru to get IPs from the cableco.
Once your router or computer is allowed to connect, it will ask for and get an IP address from the cableco, normally the same one it got the last time it asked for an address, based on the MAC making the request. If the MAC is different, you therefore get a different IP address.
Thus, to force an IP address change, you change the MAC address you are using to connect, then reboot the modem so it forgets the old one and allows the new one thru. You should then get a new IP address. Eventually, the allotted IP addresses available for assignment will be used up, and the DHCP server will start reusing ones that aren't in use. However, that typically takes quite some time, because there's a significant number of "free" IPs, and because few people cycle MAC addresses. Thus, the system is likely to remember the MAC off your old Ethernet card you used a year ago, even tho it hasn't been connected since, and still give you the same IP address, unless of course you artificially set a different MAC address.
That brings us back to where we were before, the problem of setting a different MAC address. It's easier on Linux (and I believe the various BSDs) than on MSWormOS, but it should be possible in any case, even if you have to get a new NIC to do it. A router should make
Frist psot with my new account =8^), tho I've lurked and occasionally posted AC for years.
/years/.
Your ISP is likely associating your assigned IP with your MAC address, as is quite common with DHCP, giving users that want a stable address at least a bit of stability. This seems to be particularly common with cable modem systems. Change that, and you get a new IP address. It's likely that easy. =8^)
MAC addresses are normally hard-burned into the hardware (Ethernet and the like), and are supposed to function as a GUID (globally unique ID), so changing your NIC is one way to change your MAC address. However, there's often a way to set them to something other than the hardware MAC address manually. The idea is that a network may have configured a specific association (like IP address, or even permission to connect, some ISPs register MAC addresses and won't allow unregistered addresses a connection or give them an IP address, MAC address filtering is also very common on wireless networks, for security reasons) that may be inconvenient to change when you switch NICs or the like, and the ability to manually set a MAC address allows one to set the same one they previously used.
How you change your MAC address, however, will obviously depend on what OS and/or Ethernet/other-connectivity drivers you use. Here on Gentoo, there's a network services module that combined with an app called macchanger, allows me to set my MAC at will, every time I bring up the connection on that interface. I have it set to entirely random, so if my ISP is tracking it, I might appear to be connecting with a multigigabit Internet backbone router one day, and an old 2 Mbit thinnet card or whatever the next, but that could cause problems if the ISP was relying on that for something, in which case I could just set it to randomize within my hardware type or just the specific NIC manufacturer. Anyway, with my randomized MAC address, I get a different IP address every time I reconnect (tho the computer keeps the same MAC and gets the same IP when I simply hibernate, aka suspend to disk). Otherwise, as yours, my ISP continues to hand people the same MAC associated IP for sometimes
On MSWormOS, I believe implementation is left to the Ethernet driver. If your NIC driver implements manually setting your MAC, it should have such a setting in the appropriate properties tab. Else you may be able to set it in the registry, if you know what you are doing. I stick to freedomware (see the sig) these days, so can't give you much more on that.
Routers often have an option to clone the MAC of a connected computer on the LAN to the WAN side, again, in case the ISP specifically authenticates in part by registered MAC address and won't allow changing it without calling them. Others let you set it specifically. Some, particularly those running OpenWRT or similar firmware, may allow fully randomized MAC addresses, much as I described I do with my Gentoo system above. You mentioned that switching between direct connect and using your router didn't change your assigned IP address, but it's possible that's because the router was already cloning your computer's MAC address to the WAN side, so the ISP saw no change.
So anyway, try changing your MAC address. You'll likely get a different IP address that way. If you don't, the worst that should happen is they won't let you connect, and you change it back. Of course, if they associate the assigned address with the login info (PPPoE or the like) or MAC address of the modem, you'll still connect, but changing the MAC address on your computer or router won't cause the IP address they assign you to change.
HTH,
Duncan