Even if space is an issue, articles could perhaps be ranked by importance (or perhaps categorized, and then have the categories ranked.) Then you could do stuff like keep less history for less important pages.
Exactly. I do think the internet is rapidly becoming a necessity, not a luxury. We've got ~7 billion people to keep alive on one small planet, we can't do that without a modern technological society, and that in turn needs a reliable, effective communications network.
There really are all sorts of humanitarian advantages to networking those areas that are having problems keeping people alive. For example, there was something I heard about a while back about a mobile phone network in, I think, Africa, which allowed its users to transfer credit to each other, and that was looking at diversifying that system into a general purpose money transfer facility. Then there was a programme on the BBC World Service yesterday about an open source software stack that turned mobile phones into handheld healthcare appliances.. (aha! found the (probably impermanent) link.)
On a more selfish note, I reckon there's a lot of untapped potential in less developed parts of the world. Sure, it takes a certain about of time to play "catch up" when it comes to science and technology (which is one of the reasons people get upset about outsourcing to the east - lower initial quality), but it should take a lot less time to catch up with tech than it did to invent it in the first place (think China), potentially yielding an equally sophisticated but "fresher", more enthusiastic population (tougher physical circumstances probably help provide an incentive, too.)
Heh, fair enough - and that sort of "Oh God, not again.." reaction is probably similar to the one many people get with IPv6.
I suppose on the surface, climate change seems more serious - after all, so what if the internet breaks? Humanity survived quite well for thousands of years without it, after all. But we've never had to keep ~7 billion people alive at the same time before, and we couldn't do that without a modern technological society, which in turn is underpinned by decent comms.
I guess I've now compounded my sins by escalating from climate change analogies to predicting the end of civilization if IPv6 isn't rolled out;-) But what I'm really saying is that we need a more responsible, global attitude to critical bits of 'infrastructure', whether that's packet-switched networks or the fundamentals of the biosphere.
Don't get me wrong, I'm not an ultra-leftie, I think there's plenty of room for competition in true capitalist style, but I also think there are some underlying foundations / fundamental platforms that have to be treated differently. Hopefully business will come to realise this on its own, as trying to force the issue (regulation, nationalization, etc.) rarely seems to work in the long-term.
Yeah, dual stack is fine, but I don't think that's what the OP was talking about.
Regarding your WAP - could you run OpenWRT on it, or replace it with a model that would? I picked up a WRT54GL recently, so I could play with stuff like IPv6. Someone even seems to have ported ptrtd and totd to it, so in theory it should be possible to go IPv6-only on the local network:)
IPv4-compatible addresses are deprecated, and IPv4-mapped addresses are basically only there so you can write an "IPv6 only" application and still transparently handle IPv4 connections. The actual system the app is running on still has to be dual stack.
IPv6-only hosts can't talk to IPv4-only hosts without help. As noted above, what could an IPv6-only node put in the source address of an outgoing IPv4 packet that would ensure it got to see any responses?
And and that risk of looking like I'm deliberately trying to shoot down your entire post (which I'm not!), apart from the address size, IPv6 is reasonably similar to IPv4. It tidies some stuff up, makes a few optimizations (no checksums, no fragmentation), but is still recognisably "IP."
Bingo. That's why I never understood DJB's similar little rant. You can't grow the address space and expect backward compatibility, because there's no way for a legacy node (router or host) to preserve the additional address bits (I suppose if you've got bits in another field that are always preserved and not used for anything important, you could use those. But we haven't, so that's the end of that particular story.)
You might also want to ask "technology architects" rather than "technology experts."
Some people are very good at learning the details of existing technologies, and figuring out how to mangle them to solve tomorrow's problems. Other people take a broader view and wonder how to solve next year's problems by creating new technologies. Both have their place, and there must equivalents on the "business" side of a business - people who try to foresee major economic events, the birth of whole new markets, etc. The fact that IPv6 is in many ways a "plumbing" issue (oops, made another tubes allusion) doesn't mean that long-term thinking isn't called for, even if many businesses aren't used to it in respect of (IT) infrastructure.
(Incidentally, the analogy to real architecture works quite well, I think. Sometimes "vision" is called for when creating new buildings, a whole fresh design; other times the traditional way of doing things, a design that has slowly accreted over the years, is fine.)
I've noticed recently that an awful lot of *nix based software is now supporting IPv6, either in the upstream source or added by distributions.
A lot of the demand for new addresses (and hence possibly for IPv6) will be on the smaller and / or more portable devices (phones, netbooks, set-top boxes) that often run Linux anyway.
I also note that the KDE guys are porting to Windows. I don't specifically know whether their apps generally support IPv6 already, and if so whether their Windows ports will, but I can't imagine it will be hard to add, or that it will be long before someone does.
In a nutshell, if Windows apps don't provide support, there will be workarounds. Workarounds, indeed, that might act as incentives to get people off Windows onto other, freer platforms..
Why you can't use NAT? Is this some IPv6 limitation or just that there are no IPv6 NAT routers?
NAT is not theoretically impossible with IPv6, but it's a big part of the ugliness IPv6 is meant to solve. AFAIK, there is no IPv6 NAT implementation for Linux for exactly this reason.
I use NAT for security (as a firewall) and for sharing a single external IP between my computers.
Lack of NAT doesn't mean lack of firewalling. Linux has ip6tables. You can still configure an IPv6 router / gateway to drop all incoming connections to machines inside your local network, apart from any exceptions you want to configure (e.g. Bittorrent.)
My dislike for IPv6 is that it is impossible to remember those long IP addresses.
Yes, but I guess that's unavoidable. One thing I would like to see is for IPv6 admin tools to support/etc/networks, or an equivalent. Presumably software like dnsmasq and avahi will be upgraded to cope with providing local IPv6 DNS, if they don't already (I haven't tested.)
This is a bit like saying there is no business case for doing something about climate change. Sure, I can't tell anyone that specific bits of their infrastructure are going to get wiped out by hurricanes, or that particular segments of their markets are going to be bankrupted and / or drowned by rising sea levels, but that doesn't mean it's not a good idea.
Similarly, I can't forecast what the oil price is going to do, whether it will be higher or lower in 12 months time than it is now. I don't know when we will hit peak oil, or if we've hit it already, and I don't know the exact consequences of that. But that certainly doesn't mean that looking at ways of reducing energy requirements, and alternative sources for them, isn't a good idea.
I can't say what will happen as IPv4 address scarcity hits. Will people be denied allocations outright? I doubt it. Will small blocks of addresses in random parts of the address space be auctioned to the highest bidders? Seems more likely. Will dealing with the huge routing tables caused by all those disconnected little blocks put stress on routers, causing reliability issues and more money to be spent on upgrades? Quite possibly. Will we see people rolling out multiple layers of NAT, and all sorts of ugly application-helpers? Probably. Will it be reliable? I doubt it.
Times are hard economically now, and as a result people pull their horns in and look for hard, specific reasons to justify effort and expenditure, particularly immediate, short-term reasons. But short-termism got us into the current (economic) mess in the first place. Step back, look at the big picture. Yes, it's fuzzy. That doesn't mean there aren't obvious trends, obvious problems -- and also some reasonably obvious, big-picture solutions.
To be considered a serious contender, any browser these days has got to be able to run big gobs of other people's native code. Stuffing it in a separate process where it can do only limited damage has to be a win.
Regarding (2): hopefully, the interface between the controller and the renderers is pretty "narrow", which should mean fewer bugs and make it easier to squeeze out any that are found. Plus the stuff that seems (to me) mostly likely to have issues, namely the HTML parsing and rendering, and the Javacript interpretation / compilation, are no longer part of the controlling process.
I think that's true, or at least it certainly used to be. Because fork() is always creating a clone of an existing process, page tables and the mappings in them can be created "lazily", i.e. when the new process begins to deviate from the parent.
Windows, IIRC (and I've not done Win32/NT development for ages), makes you specify either an executable file or code, data & stack sections to build the new process out of, so if they do do any optimizations, it must be much harder work.
WinXP running in Virtualbox with about 50% of the RAM.
Auto login set up on Ubuntu and WinXP, so apart from the Ubuntu splash screen, there's nothing particularly scary to see for the dyed-the-wool Windows user I'm jumping through all these hoops for.
This allows various cool stuff: incoming HTTP and IMAP connections could be scanned with ClamAV, for example. What would be really great would be to just discard changes to the main VB disk image at the end of every session. Obviously user docs + data would be somewhere else, and could potentially get infected, but that's a lot less data to periodically virus scan, or to restore if anything does get in to it.
Preliminary tests suggest that virtualized windows without on-access scanning runs quite a lot more smoothly than a bare-metal install does with it. The added bonus is that I can ssh into the underlying Ubuntu system and do admin with the rather richer toolset available there than on Windows (though greater personal familiarity with that toolset is also an issue, I admit.)
From TFA: ".. connectivity problems left some brokers unable to trade. [The LSE] was then forced to suspend trading to ensure some market players were not disadvantaged."
OTOH, scarce resource availability during "peak hours" is not exactly a new phenomenon. I pay the government to provide "road bandwidth", but my journey time through my town still doubles during the rush hour, and I doubt any amount of complaining is going to change that.
For stuff that doesn't have to be done at the same time as everything else, it only seems considerate to defer it. My ISP dislikes P2P between (IIRC) 6pm and 11pm; if you persistently hog bandwidth during those hours they move you to a separate bandwidth allocation pool which is shared with the other hogs. Outside that time, you can do what you want (lawsuit risks notwithstanding) - that seems pretty reasonable..
Another point that might be worth making: my ISP could install enough capacity, and buy sufficient transit from their upstream(s), to allow everyone to run their links at full throttle, but that wouldn't guarantee that the remote 'leaf' networks people want to access would be up to the job..
I have a feeling that behaviour might be "by design." From this blog entry:
".. select Never if you don't want to accept this upgrade offer; we might send you another offer again in the future, but it won't be for several weeks or months.."
I don't know whether your "few" matches up with Mozilla's "several":/
I haven't got one, and I got fed up with losing data when I had a power outage, or a kernel crash (some of the drivers for my hardware are a little dodgy.)
I'm back on ext3 now while I wait for ext4 to stabilize.
I do this occasionally, and it's not because I'm illiterate. When I'm writing, I tend to vocalize what I'm composing to make sure it "scans" nicely, and if I'm not concentrating the thread that actually presses keys transcribes the 'audio' in my head rather than copying the text, and so makes mistakes like this.
In most cases, it was our innovations that lead to their ability to even have a protocol to mandate.
This is one of the interesting little wrinkles about tech. The early innovators prove that something can be done, where nobody else would have tried. But those who come later get to analyze the problems of the first version, and spend more time fixing them. Think NTSC v. PAL, think Linux desktops v. Windows;-)
While places like China are, I'm sure, not terribly pleasant to live in, they may just be able to leverage this phenomenon. We've invented the science & tech., they're catching up, but their social and political structures may allow them to do it 'better.' China seems quite hot on IPv6. They are also increasingly focussed on environmental issues. They get a lot of stick on that front because of the sheer number of people, and the rate of industrial / technological expansion they're undergoing (with the related increases in pollution, CO2 emissions, etc.) But they are are already looking at solutions here, and unlike the West they can and are beginning to enforce them on a wide scale.
Not a prospect to be welcomed by individualists and libertarians, but if it shows the world that solutions on this scale are possible, the rest of us may choose to follow on voluntarily. That would be an outcome worth having..
I never quite understood that little rant. Growing the address space size will by definition break stuff, because an old client won't have a way to specify the extra bits of a new-style address, and so won't be able to get a packet to where it's going.
It's no use an IPv6 host being able to send packets an IPv4 host, if that host can't reply.
(DJB is very smart about a lot of things, and also quite obstinate about a lot of others. Something he has in common with people like Linus and Theo de Raadt. Though to be fair, having strong opinions is fairly key to actually getting stuff done.)
There are inter-op solutions, none of them particularly elegant, but hopefully they're all transient. 6to4 and Teredo give IPv6 connectivity through an IPv4 network, and TRT gives reasonable (i.e. hopefully no worse than IPv4 NAT) IPv4 connectivity to IPv6 only hosts.
Slashdot Mirror
Even if space is an issue, articles could perhaps be ranked by importance (or perhaps categorized, and then have the categories ranked.) Then you could do stuff like keep less history for less important pages.
Exactly. I do think the internet is rapidly becoming a necessity, not a luxury. We've got ~7 billion people to keep alive on one small planet, we can't do that without a modern technological society, and that in turn needs a reliable, effective communications network.
There really are all sorts of humanitarian advantages to networking those areas that are having problems keeping people alive. For example, there was something I heard about a while back about a mobile phone network in, I think, Africa, which allowed its users to transfer credit to each other, and that was looking at diversifying that system into a general purpose money transfer facility. Then there was a programme on the BBC World Service yesterday about an open source software stack that turned mobile phones into handheld healthcare appliances .. (aha! found the (probably impermanent) link.)
On a more selfish note, I reckon there's a lot of untapped potential in less developed parts of the world. Sure, it takes a certain about of time to play "catch up" when it comes to science and technology (which is one of the reasons people get upset about outsourcing to the east - lower initial quality), but it should take a lot less time to catch up with tech than it did to invent it in the first place (think China), potentially yielding an equally sophisticated but "fresher", more enthusiastic population (tougher physical circumstances probably help provide an incentive, too.)
Heh, fair enough - and that sort of "Oh God, not again .." reaction is probably similar to the one many people get with IPv6.
I suppose on the surface, climate change seems more serious - after all, so what if the internet breaks? Humanity survived quite well for thousands of years without it, after all. But we've never had to keep ~7 billion people alive at the same time before, and we couldn't do that without a modern technological society, which in turn is underpinned by decent comms.
I guess I've now compounded my sins by escalating from climate change analogies to predicting the end of civilization if IPv6 isn't rolled out ;-) But what I'm really saying is that we need a more responsible, global attitude to critical bits of 'infrastructure', whether that's packet-switched networks or the fundamentals of the biosphere.
Don't get me wrong, I'm not an ultra-leftie, I think there's plenty of room for competition in true capitalist style, but I also think there are some underlying foundations / fundamental platforms that have to be treated differently. Hopefully business will come to realise this on its own, as trying to force the issue (regulation, nationalization, etc.) rarely seems to work in the long-term.
Yeah, dual stack is fine, but I don't think that's what the OP was talking about.
Regarding your WAP - could you run OpenWRT on it, or replace it with a model that would? I picked up a WRT54GL recently, so I could play with stuff like IPv6. Someone even seems to have ported ptrtd and totd to it, so in theory it should be possible to go IPv6-only on the local network :)
IPv4-compatible addresses are deprecated, and IPv4-mapped addresses are basically only there so you can write an "IPv6 only" application and still transparently handle IPv4 connections. The actual system the app is running on still has to be dual stack.
IPv6-only hosts can't talk to IPv4-only hosts without help. As noted above, what could an IPv6-only node put in the source address of an outgoing IPv4 packet that would ensure it got to see any responses?
And and that risk of looking like I'm deliberately trying to shoot down your entire post (which I'm not!), apart from the address size, IPv6 is reasonably similar to IPv4. It tidies some stuff up, makes a few optimizations (no checksums, no fragmentation), but is still recognisably "IP."
Bingo. That's why I never understood DJB's similar little rant. You can't grow the address space and expect backward compatibility, because there's no way for a legacy node (router or host) to preserve the additional address bits (I suppose if you've got bits in another field that are always preserved and not used for anything important, you could use those. But we haven't, so that's the end of that particular story.)
You might also want to ask "technology architects" rather than "technology experts."
Some people are very good at learning the details of existing technologies, and figuring out how to mangle them to solve tomorrow's problems. Other people take a broader view and wonder how to solve next year's problems by creating new technologies. Both have their place, and there must equivalents on the "business" side of a business - people who try to foresee major economic events, the birth of whole new markets, etc. The fact that IPv6 is in many ways a "plumbing" issue (oops, made another tubes allusion) doesn't mean that long-term thinking isn't called for, even if many businesses aren't used to it in respect of (IT) infrastructure.
(Incidentally, the analogy to real architecture works quite well, I think. Sometimes "vision" is called for when creating new buildings, a whole fresh design; other times the traditional way of doing things, a design that has slowly accreted over the years, is fine.)
I've noticed recently that an awful lot of *nix based software is now supporting IPv6, either in the upstream source or added by distributions.
A lot of the demand for new addresses (and hence possibly for IPv6) will be on the smaller and / or more portable devices (phones, netbooks, set-top boxes) that often run Linux anyway.
I also note that the KDE guys are porting to Windows. I don't specifically know whether their apps generally support IPv6 already, and if so whether their Windows ports will, but I can't imagine it will be hard to add, or that it will be long before someone does.
In a nutshell, if Windows apps don't provide support, there will be workarounds. Workarounds, indeed, that might act as incentives to get people off Windows onto other, freer platforms ..
Why you can't use NAT? Is this some IPv6 limitation or just that there are no IPv6 NAT routers?
NAT is not theoretically impossible with IPv6, but it's a big part of the ugliness IPv6 is meant to solve. AFAIK, there is no IPv6 NAT implementation for Linux for exactly this reason.
I use NAT for security (as a firewall) and for sharing a single external IP between my computers.
Lack of NAT doesn't mean lack of firewalling. Linux has ip6tables. You can still configure an IPv6 router / gateway to drop all incoming connections to machines inside your local network, apart from any exceptions you want to configure (e.g. Bittorrent.)
My dislike for IPv6 is that it is impossible to remember those long IP addresses.
Yes, but I guess that's unavoidable. One thing I would like to see is for IPv6 admin tools to support /etc/networks, or an equivalent. Presumably software like dnsmasq and avahi will be upgraded to cope with providing local IPv6 DNS, if they don't already (I haven't tested.)
This is a bit like saying there is no business case for doing something about climate change. Sure, I can't tell anyone that specific bits of their infrastructure are going to get wiped out by hurricanes, or that particular segments of their markets are going to be bankrupted and / or drowned by rising sea levels, but that doesn't mean it's not a good idea.
Similarly, I can't forecast what the oil price is going to do, whether it will be higher or lower in 12 months time than it is now. I don't know when we will hit peak oil, or if we've hit it already, and I don't know the exact consequences of that. But that certainly doesn't mean that looking at ways of reducing energy requirements, and alternative sources for them, isn't a good idea.
I can't say what will happen as IPv4 address scarcity hits. Will people be denied allocations outright? I doubt it. Will small blocks of addresses in random parts of the address space be auctioned to the highest bidders? Seems more likely. Will dealing with the huge routing tables caused by all those disconnected little blocks put stress on routers, causing reliability issues and more money to be spent on upgrades? Quite possibly. Will we see people rolling out multiple layers of NAT, and all sorts of ugly application-helpers? Probably. Will it be reliable? I doubt it.
Times are hard economically now, and as a result people pull their horns in and look for hard, specific reasons to justify effort and expenditure, particularly immediate, short-term reasons. But short-termism got us into the current (economic) mess in the first place. Step back, look at the big picture. Yes, it's fuzzy. That doesn't mean there aren't obvious trends, obvious problems -- and also some reasonably obvious, big-picture solutions.
The size of the package might be a clue. I hope the original isn't 34Mb ..
There used to be an option to deal with that, but it seems to have disappeared, sadly ..
More likely to be working ball-point pens. The busted ones always seems to boomerang right back onto my desk..
I think Chrome ties the pop-ups to the creating page, and leaves them in the same process. Which isn't to say it can't be fooled ..
Regarding (1): Plugins.
To be considered a serious contender, any browser these days has got to be able to run big gobs of other people's native code. Stuffing it in a separate process where it can do only limited damage has to be a win.
Regarding (2): hopefully, the interface between the controller and the renderers is pretty "narrow", which should mean fewer bugs and make it easier to squeeze out any that are found. Plus the stuff that seems (to me) mostly likely to have issues, namely the HTML parsing and rendering, and the Javacript interpretation / compilation, are no longer part of the controlling process.
I think that's true, or at least it certainly used to be. Because fork() is always creating a clone of an existing process, page tables and the mappings in them can be created "lazily", i.e. when the new process begins to deviate from the parent.
Windows, IIRC (and I've not done Win32/NT development for ages), makes you specify either an executable file or code, data & stack sections to build the new process out of, so if they do do any optimizations, it must be much harder work.
I'm pondering the following set-up:
This allows various cool stuff: incoming HTTP and IMAP connections could be scanned with ClamAV, for example. What would be really great would be to just discard changes to the main VB disk image at the end of every session. Obviously user docs + data would be somewhere else, and could potentially get infected, but that's a lot less data to periodically virus scan, or to restore if anything does get in to it.
Preliminary tests suggest that virtualized windows without on-access scanning runs quite a lot more smoothly than a bare-metal install does with it. The added bonus is that I can ssh into the underlying Ubuntu system and do admin with the rather richer toolset available there than on Windows (though greater personal familiarity with that toolset is also an issue, I admit.)
From TFA: ".. connectivity problems left some brokers unable to trade. [The LSE] was then forced to suspend trading to ensure some market players were not disadvantaged."
OTOH, scarce resource availability during "peak hours" is not exactly a new phenomenon. I pay the government to provide "road bandwidth", but my journey time through my town still doubles during the rush hour, and I doubt any amount of complaining is going to change that.
For stuff that doesn't have to be done at the same time as everything else, it only seems considerate to defer it. My ISP dislikes P2P between (IIRC) 6pm and 11pm; if you persistently hog bandwidth during those hours they move you to a separate bandwidth allocation pool which is shared with the other hogs. Outside that time, you can do what you want (lawsuit risks notwithstanding) - that seems pretty reasonable ..
Another point that might be worth making: my ISP could install enough capacity, and buy sufficient transit from their upstream(s), to allow everyone to run their links at full throttle, but that wouldn't guarantee that the remote 'leaf' networks people want to access would be up to the job..
I have a feeling that behaviour might be "by design." From this blog entry:
I don't know whether your "few" matches up with Mozilla's "several" :/
If you're running XFS, you certainly do want a UPS.
I haven't got one, and I got fed up with losing data when I had a power outage, or a kernel crash (some of the drivers for my hardware are a little dodgy.)
I'm back on ext3 now while I wait for ext4 to stabilize.
I do this occasionally, and it's not because I'm illiterate. When I'm writing, I tend to vocalize what I'm composing to make sure it "scans" nicely, and if I'm not concentrating the thread that actually presses keys transcribes the 'audio' in my head rather than copying the text, and so makes mistakes like this.
And yes, I know I have a weird brain :)
In most cases, it was our innovations that lead to their ability to even have a protocol to mandate.
This is one of the interesting little wrinkles about tech. The early innovators prove that something can be done, where nobody else would have tried. But those who come later get to analyze the problems of the first version, and spend more time fixing them. Think NTSC v. PAL, think Linux desktops v. Windows ;-)
While places like China are, I'm sure, not terribly pleasant to live in, they may just be able to leverage this phenomenon. We've invented the science & tech., they're catching up, but their social and political structures may allow them to do it 'better.' China seems quite hot on IPv6. They are also increasingly focussed on environmental issues. They get a lot of stick on that front because of the sheer number of people, and the rate of industrial / technological expansion they're undergoing (with the related increases in pollution, CO2 emissions, etc.) But they are are already looking at solutions here, and unlike the West they can and are beginning to enforce them on a wide scale.
Not a prospect to be welcomed by individualists and libertarians, but if it shows the world that solutions on this scale are possible, the rest of us may choose to follow on voluntarily. That would be an outcome worth having ..
I never quite understood that little rant. Growing the address space size will by definition break stuff, because an old client won't have a way to specify the extra bits of a new-style address, and so won't be able to get a packet to where it's going.
It's no use an IPv6 host being able to send packets an IPv4 host, if that host can't reply.
(DJB is very smart about a lot of things, and also quite obstinate about a lot of others. Something he has in common with people like Linus and Theo de Raadt. Though to be fair, having strong opinions is fairly key to actually getting stuff done.)
There are inter-op solutions, none of them particularly elegant, but hopefully they're all transient. 6to4 and Teredo give IPv6 connectivity through an IPv4 network, and TRT gives reasonable (i.e. hopefully no worse than IPv4 NAT) IPv4 connectivity to IPv6 only hosts.
Nice one, thanks. I'm not good at processing information that I can't hop around in and scan these days (I think it's age setting in.)