I used to have a TV, but stopped watching when I was writing up my phd and never really got back in to it. When the renewal came round I realised I'd watched maybe half a dozen programmes in a year and so decided not to renew. A bit later I got the expected "Are you sure you've not got a TV?" letter, which I replied to. I've since heard nothing. I think that was a year ago February. So occasionally TVL aren't complete pains in the bum.
You're right. The ipv4 address report at potaroo is a prediction based on modelling and it does change. A while back I started recording the reports and plotting the changes in predictions. It's a bit disappointing that I didn't start before the world began to end because I bet the graph would be a much more interesting shape. Anyway, current predictioned date are getting further away - the number of days remaining at the time the report is made remains roughly constant.
ARIN recovered a/8 in 2007. It's unlikely they'll get any more back. I know that doesn't sound much, but the amount of effort involved in getting address space back means that it is probably not worth it. Who pays for getting the company to move their addresses? How long will it take? I wouldn't be surprised if the legal wrangling took a long time to sort things out. Given that we're using about one/8 per month, it won't help that much even if we could get a few blocks back.
Take a read of http://www.potaroo.net/tools/ipv4/index.html to see the rate at which ipv4 addresses are being allocated, along with their predictions for the future. There's a lot there, but it's worth reading at least a bit of it:)
A while back, I wondered how their prediction changed over time so started logging it. The results of that are at http://atchoo.org/ipv4/
Yes, I do want access to the desktops. If I have to, I'll try VNC, but my experience of it has been that it's horribly slow to use, even over 100Mb. I can use Dameware to control machines of users that have dialed in from home, and it's still more than quick enough. It also has nice features like auto-reconnect that'll keep pinging a rebooting machine and automatically connect back up to it as soon as it's available.
I'm still not sure we're talking about the same usage...:) NX gives you the ability to have a graphical desktop much like Windows Remote Desktop. Any number of people can use it at once on the same machine. It does not give you the ability to control the session of an already logged in user in the same way as Remote Assistance (iirc) works. I don't know about dialup use, but I've used NX over low hundreds of k/s connections without any bother.
I think you're stuck on a lot of those points, alas. This are my answers to selected questions, bearing in mind that I'm not a proper admin, even though I do look after 20 machines or so at work.
> I want to enforce the installation of updates
You could add a cron entry for root to install updates at a specified time each day and grab the updates from the repository that you run. Your repository can be in addition to the normal distro repositories, so you don't need to replicate everything.
> **Group Policy Software Deployment**
I don't think there is anything exactly as you describe, but there are tools that allow running of commands at multiple machines in parallel - I've never looked into this myself, but I believe they might be "cluster ssh" or "parallel ssh" or something similar. This bit is pure speculation, but I'd imagine that one of those, coupled with adding/removing machines to netgroups would look vaguely like what you describe.
> **Roaming Home Folders**
I imagine you're going to get lots of replies here saying "just use nfs!", and I'd be inclined to go with that myself. The usual way this is handled in my experience is to have the home directories on a remote server and that auto mounted with nfs when you log on to a machine. This sounds like what you describe apart from the part with a profile being copied to the local machine. I'm not sure I understand the benefit of that, could you clarify?
> **Preventing access to Executables**
As others have mentioned, you can mount anything with the "noexec" flag and nothing can be executed on that partition/network mount (is that what you meant by removing the execute flag? I wasn't sure whether you meant the execute flag on the file itself). As you're the only person with the ability to install software, bingo, they can only run what you want. If you want more fine grained control, looking into something like selinux would do it, along with a significant amount of hassle.
> **Remote Support**
Agreed, VNC is a bit clunky. I use nomachine nx for remote access and it's the bees knees. This doesn't let you connect to an existing user session if that's what you mean by "remote access to any users desktop" though.
> Do you really need separate partitions for all these things?
On a user machine, not really. It can be useful at times though. Our machines at work typically have 20GB root partition for the OS, ?GB of swap and then the rest is set aside for installs of the *big* commercial apps that we use and temporary simulation data and the like. This partition is preserved during OS upgrades, so we don't have to reinstall which is nice. Essentially I'm saying 20GB (or more if you want, obviously it won't hurt too much to increase that) and an amount for swap is all you need, assuming your home directories are on the network.
I hope that answers some of your questions. I'll be interested to see what other people write as well!
hg has both a revision number and a changeset id. The revision number is human readable and useful within a particular repo. The changeset id is unique across repos, the same as git does.
I think it's got a fixed time limit rather than a distance limit. This should mean that it selects for speed as well as ability to cross the terrain. Pretty neat either way!
Generating small numbers of photons is quite commonplace. Fluorescence imaging is often done as single photon or two photon imaging - there's lots if you search for that in google (and pictures too!). It's exactly the kind of thing you describe - excite the atom with your incoming light pulse and then get a different wavelength out when the electons decays to its previous state. You'll likely be looking at using femto-second lasers (ie. pulses a fs long) to do it. I confess this is an area that I'm not particularly familiar with. Don't forget that you might be spreading a reasonable few photons over a large area as well, so although the total number of photons could be relatively high, the number per pixel detector could still be low.
Avalanche photodiodes (at least the ones that I'm working with) are simply silicon diodes biased past their breakdown voltage. This means that there is a strong field across the junction and essentially any charge crossing the junction will cause it to go into breakdown, ie. pass a lot of current and reduce the voltage across the diode dramatically. This is exactly what happens when a photon is aborbed and creates an electron-hole pair. Similarly, it also happens due to thermally generated electrons. The really nice thing about the voltage signal coming out of the diode is that it's essentially just a digital edge, so there's no need for analogue-to-digital conversion - you just know that a photon has appeared. The problem, as this article mentions, is that when your diode is in breakdown, there's (normally) no way to detect any further photons because the diode hasn't recovered.
What gets me still is the wave-particle duality of light. Working with photodiodes, I treat light exclusively as photons and everything else just makes my head hurt:)
I mirror the back catalogue and have no intention of getting rid of it. I know other mirror maintainers have said the same thing. Talking to Aq on irc today he gave the impression that the main archive will be sticking around as well, so don't worry too much:)
KDE guys don't guarantee that KDE 3.5 will be supported next 3 years
Really? That sounds horrible, the development model of KDE sounds very amateurish.
I'm not disagreeing with you, but can you provide examples of open source projects that *do* guarantee to provide support for versions of their code that are three years old? As a reference, for GNOME this would be 2.8 and the Linux kernel would be 2.6.10.
I guess articles on the FreeBSD schedulers are either not being submitted or are being rejected. It's a shame either way. I don't use BSD (for no particular reason) but I'd still very much like to hear about what's going on there.
Thus far, we have about a century's worth of data, but even then, the early data is susceptible to more errors and/or not complete. So, probably somewhere around 2250 can we start talking about real Global Warming and have enough data that most will actually believe it.
How much would you trust data from 250 years ago? I'll bet that in 250 years there will be plenty of doubt as to the accuracy of our current measurements.
Has it crossed anyone's mind that we could be alone? Bacteria and other forms of life aside, is there any reason to believe that galaxies team with life, just because there are so many?
Something like 90% of the solar systems out there contain *nothing* but gasses, a bunch more have planets too far or too close to their sun. Still more are in flux- near black holes, and pulsars and the like.
I'd love to see something out there, too, but I'm a little dubious about the likelihood. There are perhaps a million combinations of circumstances that allow us to be, and be here. Everything from a precise gravity to an orbiting moon; it's not a simple circumstance.
Very true, but I think the point is that there are staggering numbers involved, so the probability that we're the only form of life is incredibly small. According to "The Universe within 14 billion Light Years", there are 350 billion large galaxies and 7 trillion dwarf galaxies in the visible universe, comprised of 30 billion trillion stars (3e22). Even when you start being pretty pessimistic about the conditions required for life you'll still end up with more than one planet being suitable.
> And the headline and summary of the slashdot article do not > do it justice. The editors should be spanked for that one
Yeah, absolutely. Hence my suggestion:) CmdrTaco posted a dupe of it a while later that disappeared quickly, but it *did* have a proper title and summary.
Slashdot Mirror
I used to have a TV, but stopped watching when I was writing up my phd and never really got back in to it. When the renewal came round I realised I'd watched maybe half a dozen programmes in a year and so decided not to renew. A bit later I got the expected "Are you sure you've not got a TV?" letter, which I replied to. I've since heard nothing. I think that was a year ago February. So occasionally TVL aren't complete pains in the bum.
Do you have any better hostages?
You're right. The ipv4 address report at potaroo is a prediction based on modelling and it does change. A while back I started recording the reports and plotting the changes in predictions. It's a bit disappointing that I didn't start before the world began to end because I bet the graph would be a much more interesting shape. Anyway, current predictioned date are getting further away - the number of days remaining at the time the report is made remains roughly constant.
Graphs at http://atchoo.org/ipv4/
http://blog.icann.org/2008/02/recovering-ipv4-address-space/
ARIN recovered a /8 in 2007. It's unlikely they'll get any more back. I know that doesn't sound much, but the amount of effort involved in getting address space back means that it is probably not worth it. Who pays for getting the company to move their addresses? How long will it take? I wouldn't be surprised if the legal wrangling took a long time to sort things out. Given that we're using about one /8 per month, it won't help that much even if we could get a few blocks back.
Take a read of http://www.potaroo.net/tools/ipv4/index.html to see the rate at which ipv4 addresses are being allocated, along with their predictions for the future. There's a lot there, but it's worth reading at least a bit of it :)
A while back, I wondered how their prediction changed over time so started logging it. The results of that are at http://atchoo.org/ipv4/
I'm still not sure we're talking about the same usage... :) NX gives you the ability to have a graphical desktop much like Windows Remote Desktop. Any number of people can use it at once on the same machine. It does not give you the ability to control the session of an already logged in user in the same way as Remote Assistance (iirc) works. I don't know about dialup use, but I've used NX over low hundreds of k/s connections without any bother.
Cheers,
Roger
I think you're stuck on a lot of those points, alas. This are my answers to selected questions, bearing in mind that I'm not a proper admin, even though I do look after 20 machines or so at work.
> I want to enforce the installation of updates
You could add a cron entry for root to install updates at a specified time each day and grab the updates from the repository that you run. Your repository can be in addition to the normal distro repositories, so you don't need to replicate everything.
> **Group Policy Software Deployment**
I don't think there is anything exactly as you describe, but there are tools that allow running of commands at multiple machines in parallel - I've never looked into this myself, but I believe they might be "cluster ssh" or "parallel ssh" or something similar. This bit is pure speculation, but I'd imagine that one of those, coupled with adding/removing machines to netgroups would look vaguely like what you describe.
> **Roaming Home Folders**
I imagine you're going to get lots of replies here saying "just use nfs!", and I'd be inclined to go with that myself. The usual way this is handled in my experience is to have the home directories on a remote server and that auto mounted with nfs when you log on to a machine. This sounds like what you describe apart from the part with a profile being copied to the local machine. I'm not sure I understand the benefit of that, could you clarify?
> **Preventing access to Executables**
As others have mentioned, you can mount anything with the "noexec" flag and nothing can be executed on that partition/network mount (is that what you meant by removing the execute flag? I wasn't sure whether you meant the execute flag on the file itself). As you're the only person with the ability to install software, bingo, they can only run what you want. If you want more fine grained control, looking into something like selinux would do it, along with a significant amount of hassle.
> **Remote Support**
Agreed, VNC is a bit clunky. I use nomachine nx for remote access and it's the bees knees. This doesn't let you connect to an existing user session if that's what you mean by "remote access to any users desktop" though.
> Do you really need separate partitions for all these things?
On a user machine, not really. It can be useful at times though. Our machines at work typically have 20GB root partition for the OS, ?GB of swap and then the rest is set aside for installs of the *big* commercial apps that we use and temporary simulation data and the like. This partition is preserved during OS upgrades, so we don't have to reinstall which is nice. Essentially I'm saying 20GB (or more if you want, obviously it won't hurt too much to increase that) and an amount for swap is all you need, assuming your home directories are on the network.
I hope that answers some of your questions. I'll be interested to see what other people write as well!
Cheers,
Roger
How about a link for those of us that do have matlab? I like unbelievably cool things :)
Someone may be inclined to port it to something more portable as well (although that's unlikely to be me at this point).
hg has both a revision number and a changeset id. The revision number is human readable and useful within a particular repo. The changeset id is unique across repos, the same as git does.
Hmm, I don't get that on any of the Linux / Solaris machines I've just tried.
Trying 127.0.0.1...
Connected to localhost.
Escape character is '^]'.
SSH-2.0-OpenSSH_5.1
I've also had 5.0 and 4.7.
I think it's got a fixed time limit rather than a distance limit. This should mean that it selects for speed as well as ability to cross the terrain. Pretty neat either way!
Thanks for the clarification. That makes much more sense than the muddled idea I had of it.
Generating small numbers of photons is quite commonplace. Fluorescence imaging is often done as single photon or two photon imaging - there's lots if you search for that in google (and pictures too!). It's exactly the kind of thing you describe - excite the atom with your incoming light pulse and then get a different wavelength out when the electons decays to its previous state. You'll likely be looking at using femto-second lasers (ie. pulses a fs long) to do it. I confess this is an area that I'm not particularly familiar with. Don't forget that you might be spreading a reasonable few photons over a large area as well, so although the total number of photons could be relatively high, the number per pixel detector could still be low.
Avalanche photodiodes (at least the ones that I'm working with) are simply silicon diodes biased past their breakdown voltage. This means that there is a strong field across the junction and essentially any charge crossing the junction will cause it to go into breakdown, ie. pass a lot of current and reduce the voltage across the diode dramatically. This is exactly what happens when a photon is aborbed and creates an electron-hole pair. Similarly, it also happens due to thermally generated electrons. The really nice thing about the voltage signal coming out of the diode is that it's essentially just a digital edge, so there's no need for analogue-to-digital conversion - you just know that a photon has appeared. The problem, as this article mentions, is that when your diode is in breakdown, there's (normally) no way to detect any further photons because the diode hasn't recovered.
What gets me still is the wave-particle duality of light. Working with photodiodes, I treat light exclusively as photons and everything else just makes my head hurt :)
Don't forget that you still have the chance to get some of the LugRadio vibe at LRL UK in July.
Details are at http://lugradio.org/live/UK2008/
Roger
I mirror the back catalogue and have no intention of getting rid of it. I know other mirror maintainers have said the same thing. Talking to Aq on irc today he gave the impression that the main archive will be sticking around as well, so don't worry too much :)
Roger
I'm sure that one could also argue that a difference in the number of hits for tutorial is indicative of the relative difficulty of the vcs :)
Cheers,
Roger
I'm not disagreeing with you, but can you provide examples of open source projects that *do* guarantee to provide support for versions of their code that are three years old? As a reference, for GNOME this would be 2.8 and the Linux kernel would be 2.6.10.
Cheers,
Roger
Thanks, saved me writing the same thing :) Flexibility is the key point here.
http://en.wikipedia.org/wiki/Asymptote ?
I guess articles on the FreeBSD schedulers are either not being submitted or are being rejected. It's a shame either way. I don't use BSD (for no particular reason) but I'd still very much like to hear about what's going on there.
> 1000Hz ticks
Don't forget that the "tickless" option is available now.
Cheers,
Roger
How much would you trust data from 250 years ago? I'll bet that in 250 years there will be plenty of doubt as to the accuracy of our current measurements.
Cheers,
Roger
The other case where you need to escape spaces is if you use them as delimeters. Only useful to confuse people of course :)
sed 's "of\ a\ pity" foreseeable g' somefile
Very true, but I think the point is that there are staggering numbers involved, so the probability that we're the only form of life is incredibly small. According to "The Universe within 14 billion Light Years", there are 350 billion large galaxies and 7 trillion dwarf galaxies in the visible universe, comprised of 30 billion trillion stars (3e22). Even when you start being pretty pessimistic about the conditions required for life you'll still end up with more than one planet being suitable.
Cheers,
Roger
> And the headline and summary of the slashdot article do not
:) CmdrTaco posted a dupe of it a while later that disappeared quickly, but it *did* have a proper title and summary.
> do it justice. The editors should be spanked for that one
Yeah, absolutely. Hence my suggestion
Cheers,
Roger