The obvious problem being that a lot of ISPs these days (especially UK-based ISPs) are blocking or severely throttling torrent traffic. It makes it rather slow to grab the latest Fedora ISOs...
The obvious solution to this problem is to provide a HTTP or FTP server to grab it from, but the catch is how are you going to get that multi-gigabyte file onto the server in the first place, unless both you and the server have got an uberfast connection?
Dedicated servers aren't the most expensive things in the universe (£80pm for a 2.4GHz server with 512MB RAM, 80GB HDD, and a 10Mbit direct pipe to the Internet last time I checked) but they can be a bit of a pig to admin...
Ah, but it's the scope for naughtiness that makes it fun. Ettercap, anyone?:)
Anyone who gives a stuff about security on WiFi should have an active firewall, and SSH to a known-secure machine (PuTTY can do Dynamic Port Forwarding, i.e. SOCKS, which beats setting up a HTTP proxy on said machine). Sniffing packets only works if those packets aren't encrypted...
LinkSys NSLU2 + Alcatel SpeedTouch 330 DSL modem + hard drive. About 5.5W power consumption for the NSLU2 and modem, plus 15W for the hard drive. Total of 20.5W (ish), and it does:
DHCP with DNS caching (dnsmasq)
Samba with NT domain serving
Mail download, filtering and dispatch (Postfix, UW-IMAP, iPOP3d, Fetchmail, Procmail, and my own homebrew spam filter)
Print server (CUPS, Rawprintd and a modified version of Epsc70stat to monitor the ink levels on my Epson Stylus C64)
All this on a server that has 32MB of RAM, and a 202MHz Xscale CPU. Took about a week to get everything working, but it was fun (and that's the important thing!):)
Rawprintd is neat - it lets you make the NSLU emulate a HP JetDirect, with as many ports as you want. It's supported natively by Win2k and XP, or you can download the HP Port Monitor Driver to use it with 95 and 98.
A few of the newer Nokia cellphone battery packs actually have holograms on them. The idea is, if the hologram is there (and it looks sharp, well printed, etc) then there's a pretty good chance the battery is authentic. Problem is, not everyone checks the hologram, so in that respect a "magic handshake" is a good plan. Stop the phone (or whatever) from booting, and make it display a "Counterfeit battery is inserted, please replace with OEM battery" warning when you try and charge it.
I seem to recall Sony's InfoLithium camcorders do a handshake/authentication check with the battery and if it responds wrong (or just doesn't respond at all) then the camera won't power up. Of course, that didn't stop the aftermarket cell and charger manufacturers from reverse engineering the protocol. Sony batteries are damned expensive...
What would be nice is if a few of the big-name manufacturers got together and had a website you could use to validate your battery. "Go to validateyourbattery.com, enter your battery serial number at the prompt and find out if it's genuine, and if so when it was manufactured". Maybe put two codes on the battery - a serial number and a validation key. You call, they tell you what the validation key should be, and if it matches then there's a decent chance the battery is genuine.
Hell, let's go the whole hog - have it tell you what laptops the battery works in too. Useful if you're given a battery by a friend, it doesn't fit your machine, and you want to know what it works with so you can sell it on Ebay.
I might actually see if that's a workable idea.. you'd have to tie the manufacturing/distribution data to the website, but a nightly update would solve that. Most DBMS software should be able to handle the data involved (I'd probably use PostgreSQL) so your only problem would be keeping the data in sync. That said, it takes a fair bit longer than 24 hours for a battery to get from Malaysia or wherever to the UK or USA, so a bit of out-of-date data wouldn't be a massive problem.
To be honest, I'm surprised it hasn't already been done...
Ah, it's got Speedstep. That might be why it runs cooler... My SatPro actually ramps down the frontside bus - it goes down as low as 8MHz IIRC. The CPU runs slower and cooler, but as a side effect the machine grinds to a complete and total standstill because of the low FSB. When it's stepped down to 200MHz CPU clock, a 200MHz Pentium MMX can actually outperform the Toshiba's Celeron.
I've had to replace the TFT (a couple of incredibly annoying dead pixels and a hot subpixel - now it's down to one green hot subpixel), a hard drive and the audio riser card. Also had to resolder the speaker filters on the mainboard. I actually dropped it (once) - the DVD drive faceplate popped off and the HDD died, but that was it as far as damage goes. I was going to upgrade the HDD anyway, and there was no un-backed-up data on the machine at the time, so nothing lost except a bit of time swapping the HDD out. A 40GB Seagate Momentus makes a very nice drive for a laptop of this spec.
The only other problem I had was that the sound would randomly cut out. There are two dark grey filter cubes on the motherboard near the audio riser card - resoldering them fixed the problem.
It's a lovely little machine - a bit big for using on the train (unless you're sitting at a table), but fantastic for getting work done at university when the computer lab is full.
I've got a Toshiba Satellite Pro 4600, which goes from "quite cool" to "OMG, my nuts are roasting!" in about an hour, less if it's running Linux with KDE, a lot longer if it's running Linux in console mode (i.e. no Xwin). That machine is a 700MHz Coppermine Celeron. It also gets about two and a half hours runtime off a single charge, which usually means that when the Uberlaptop Brigade have had their batteries die, mine usually has at least an hour of runtime left. I've got the Duracell Bunny of laptops:)
Just out of curiosity, failure-man, what machine are you using?
Must have been a pretty catastrophic failure. There are usually at least three levels of protection on any lithium-ion battery:
PTC cutoff switch - there's an overtemperature cutoff switch inside each Li-ion cell. If the cell gets too hot, the resettable cutoff switch disconnects the battery from the others. If it gets really, REALLY hot, then that cutoff becomes permanent. This is a last resort protection device, and should NOT be relied on to work. It's there to try and limit damage if everything else fails.
Protection circuit - monitors state-of-charge, voltage, temperature and input/output current of the battery. If the battery voltage is too low or too high, or if you try to overcharge or over discharge the pack, the circuit disconnects the battery pack from everything else until the voltage, etc. returns to its normal 'safe' range. If you REALLY piss off the protection circuit, it will literally blow a fuse - most packs are fitted with an "SC-Protector", which is basically a fuse that can be blown by an electrical signal (a "self destruct input" if you will). Not only do you get the overcurrent characteristics of a fuse, you also get to blow it if something bad happens. SC-Protectors are not readily available in quantities less than a thousand or so, so once it's blown your battery pack becomes a brick.
Charge controller - the charging circuit should continuously monitor its output, and shut down if it goes out of range. Not all of them do...
The problem with Li-ion (and to a greater degree, Lithium Polymer) cells is that they're so sensitive - charge them over 4.2V or discharge them below 3.2V and the cell will be damaged. Abuse it a lot and it will blow up. To get that to happen in a properly designed circuit, you'd need a chain of failures:
First, the protection circuit would have to fail in some way that would prevent it from protecting the battery pack. A shorted switching transistor (usually a MOSFET) and a dead SC-Protector drive transistor would do that nicely. The protection IC can see something's really, really wrong, but it won't be able to do anything about it. Bear in mind the switching transistor has to handle the power of charging and discharging - it takes quite a beating. Shorted MOSFETs really aren't that rare.
Next you'd have to have a failure in the charge circuit that causes the battery to be overcharged. For the sake of argument, let's say the charge IC has latched up. It no longer regulates its output voltage properly and - again, for the sake of argument - we'll say that there's 5V over each cell instead of 4.2V.
Now that fault condition has to exist for long enough that the cell electrolyte will break down (usually into hydrogen and other miscellaneous nasties). When that happens, the safety vent will fire and the battery ejects a hot stream of gas.
Now the final act. A little spark (say, from the PTC switch) and that hydrogen catches fire. That flame heats up another cell to the point where the vent fires, and you have a second cell joining in the explosive fun. In a few seconds, that cell will set fire to another, and the process will keep on repeating until the cells burn each other out, and the fire runs out of fuel.
The big problem with Li-ions is that they're inherently unstable. The nickel-based batteries tend to be much more forgiving of abuse. They usually don't blow up unless you really, really abuse them. You might damage them and reduce their capacity a bit, but you usually won't be able to make them explode or spontaneously combust without some serious work. They do have a lower energy density and terminal voltage than Li-ion and Li-Polymer, though, which might partially explain why they're more stable.
You don't like the large icons? "Buddies" menu, "Show Buddy Details" off, and "Show Empty Groups" off if you want to (potentially) get rid of a bit more clutter. The procedure might be a bit different on older versions, but that's how you do it on the 2.0 betas.
Slashdot Mirror
The obvious solution to this problem is to provide a HTTP or FTP server to grab it from, but the catch is how are you going to get that multi-gigabyte file onto the server in the first place, unless both you and the server have got an uberfast connection?
Dedicated servers aren't the most expensive things in the universe (£80pm for a 2.4GHz server with 512MB RAM, 80GB HDD, and a 10Mbit direct pipe to the Internet last time I checked) but they can be a bit of a pig to admin...
Anyone who gives a stuff about security on WiFi should have an active firewall, and SSH to a known-secure machine (PuTTY can do Dynamic Port Forwarding, i.e. SOCKS, which beats setting up a HTTP proxy on said machine). Sniffing packets only works if those packets aren't encrypted...
That friend is me :)
- DHCP with DNS caching (dnsmasq)
- Samba with NT domain serving
- Mail download, filtering and dispatch (Postfix, UW-IMAP, iPOP3d, Fetchmail, Procmail, and my own homebrew spam filter)
- Web serving (Cherokee)
- Subversion version control
- Print server (CUPS, Rawprintd and a modified version of Epsc70stat to monitor the ink levels on my Epson Stylus C64)
All this on a server that has 32MB of RAM, and a 202MHz Xscale CPU. Took about a week to get everything working, but it was fun (and that's the important thing!)Rawprintd is neat - it lets you make the NSLU emulate a HP JetDirect, with as many ports as you want. It's supported natively by Win2k and XP, or you can download the HP Port Monitor Driver to use it with 95 and 98.
They also use a heck of a lot less power than a full PC. Its power supply is 12 volts at half an amp, which works out at about six Watts, worst-case.
I seem to recall Sony's InfoLithium camcorders do a handshake/authentication check with the battery and if it responds wrong (or just doesn't respond at all) then the camera won't power up. Of course, that didn't stop the aftermarket cell and charger manufacturers from reverse engineering the protocol. Sony batteries are damned expensive...
What would be nice is if a few of the big-name manufacturers got together and had a website you could use to validate your battery. "Go to validateyourbattery.com, enter your battery serial number at the prompt and find out if it's genuine, and if so when it was manufactured". Maybe put two codes on the battery - a serial number and a validation key. You call, they tell you what the validation key should be, and if it matches then there's a decent chance the battery is genuine.
Hell, let's go the whole hog - have it tell you what laptops the battery works in too. Useful if you're given a battery by a friend, it doesn't fit your machine, and you want to know what it works with so you can sell it on Ebay.
I might actually see if that's a workable idea.. you'd have to tie the manufacturing/distribution data to the website, but a nightly update would solve that. Most DBMS software should be able to handle the data involved (I'd probably use PostgreSQL) so your only problem would be keeping the data in sync. That said, it takes a fair bit longer than 24 hours for a battery to get from Malaysia or wherever to the UK or USA, so a bit of out-of-date data wouldn't be a massive problem.
To be honest, I'm surprised it hasn't already been done...
I've had to replace the TFT (a couple of incredibly annoying dead pixels and a hot subpixel - now it's down to one green hot subpixel), a hard drive and the audio riser card. Also had to resolder the speaker filters on the mainboard. I actually dropped it (once) - the DVD drive faceplate popped off and the HDD died, but that was it as far as damage goes. I was going to upgrade the HDD anyway, and there was no un-backed-up data on the machine at the time, so nothing lost except a bit of time swapping the HDD out. A 40GB Seagate Momentus makes a very nice drive for a laptop of this spec.
The only other problem I had was that the sound would randomly cut out. There are two dark grey filter cubes on the motherboard near the audio riser card - resoldering them fixed the problem.
It's a lovely little machine - a bit big for using on the train (unless you're sitting at a table), but fantastic for getting work done at university when the computer lab is full.
Just out of curiosity, failure-man, what machine are you using?
The problem with Li-ion (and to a greater degree, Lithium Polymer) cells is that they're so sensitive - charge them over 4.2V or discharge them below 3.2V and the cell will be damaged. Abuse it a lot and it will blow up. To get that to happen in a properly designed circuit, you'd need a chain of failures:
The big problem with Li-ions is that they're inherently unstable. The nickel-based batteries tend to be much more forgiving of abuse. They usually don't blow up unless you really, really abuse them. You might damage them and reduce their capacity a bit, but you usually won't be able to make them explode or spontaneously combust without some serious work. They do have a lower energy density and terminal voltage than Li-ion and Li-Polymer, though, which might partially explain why they're more stable.
You don't like the large icons? "Buddies" menu, "Show Buddy Details" off, and "Show Empty Groups" off if you want to (potentially) get rid of a bit more clutter. The procedure might be a bit different on older versions, but that's how you do it on the 2.0 betas.
> is 522 weeks assuming 635.25 days/year.
Hey, maybe he RSA encrypted it - so all we need to do is find the right values of N and D and... Oh, wait, he posted a correction. Never mind.