Almost all thermal plants, regardless of heat source, use a closed steam loop. Otherwise, you have to continually de-mineralize the feedwater, and deal with precipitate in the boiler.
How the steam is condensed varies; if you're near enough water, you can just use that. If you're near enough to a town, you can sink your waste heat into the district heating system. If you're out of all other options, you can build one of those hyperboloid cooling towers people associate with nuclear plants. (They're actually used with all sorts of thermal plants, including coal and oil.)
So, even the little 550 MW combined-cycle gas-fired plant in my neighbourhood recycles steam. (Combined-cycle means the waste heat from gas turbines boils water to run a steam turbine.)
Right, and there's prior art using soundwaves as memory: mercury delay-lines.
Just don't close the loop and, instead, you have a point-to-point sound transmission system.
In the Real World, that nasty place that makes all our math hard, you'd have to deal with reflections and "impedance mismatches" (where pipes of two different sizes meet). How it would work if you went with an Aloha-net style multiple transceivers on a shared medium I'm not sure, but... I could see a decent RFC coming out of this for the day after March.
Not so much; this is more danger approaching from the side. Generally in situations where the pedestrian has right-of-way--like his example of crossing the street and someone making a left turn.
If you can hear the car coming, you can get out of the way--even when it's the car driver who should be yielding. You're still hurt even when you're not at fault.
We could ask for drivers to be held to the actual traffic law, but history is against us, here.
To further clarify the mlts' response, you hook up ONLY transmit lines on the transmitter's side. You leave out all of the handshake lines going the other way, so no RTS/CTS handshake; definitely no XON/XOFF.
If the transmitter is too fast for the receiver, the receiver will buffer-overrun and corrupt the data it sees; the UART hardware SHOULD set a status flag when it overruns. But there MUST NOT be any way for the receiver to tell the transmitter to slow down.
It is acceptable to have lines from the transmitter that tell the receiver it is "online", like hooking DTR to DSR. But it can't be done the other way; the transmitter just sends blind.
The usual way of dealing with errors in this sort of situation is to send the data multiple times. That means each message needs a sequence number. (And don't forget to include the sequence number in the checksum; so you drop the whole message if the checksum fails.)
The first step in reducing the error rate, though, is reducing the speed. RS-232 (and -422 and -423) are well-understood and quite robust--in part because they don't use really high speeds, and their data clock is much slower than their sampling clock--a nature of the asynchronous beast.
If you were really going to do this, say for modern boxes with USB, you can even get USB-to-RS-232 cables that end in pigtails or sockets and use TTL voltages; you don't need to bother with the +/-12V level conversion. Just wire RX to TX, GND to GND, and you're ready. (Bonus points for hooking DTR to DCD or something like that for detecting "remote live, no transmission".)
And if I was doing this "for real", I'd use optoisolators so the two boxes really aren't electrically connected: they don't even need to share a ground. Like the way MIDI serial works; it's an electrical current-loop like RS-232, but the receiver is an optoisolator so you don't get ground loops between your instruments and sequencer. (Once you get past that, it's basically RS-232 with an unusual clock oscillator; you can interface to it with standard UART ASICs.)
Errr. So basically what I'd do is use a communication system that is exactly like MIDI electrically (PHY, layer 1) but put my on protocol on it.
Actually, if you pull up the same chart on Yahoo! Finance and switch it to linear scale, it looks much, much worse for Ballmer. With a linear scale, it looks like they were going to hit +INF under Gates, and it immediately plummets under Ballmer.
(And for the AC replying to you, 0.01 displayed to 1 decimal place is 0.0.)
Only problem is, underscore isn't allowed for most DNS records, especially A, AAAA and CNAME. You'll have to get anonymous-coward.xxx. (Hostnames within it are your own business, if you feel you need www.)
Him: "We're just going to start in the Cloud, we won't bother with a virtual server to start off."
*hands over server information*
Me: "Hmmm, that looks exactly like a virtual server with 'virtual server' crossed out and 'Cloud' written in with a crayon. Right down to where it uses Xen."
I've easily got 30-40 windows open at once. Console sessions on a dozen remote machines, several Web browsers with distractions and reference manuals, e-mail window, ticketing system, VMware GUI (only needed for Windows VMs, all the UNIX/Linux/BSD ones can be got to via SSH from a native terminal), a few Emacs "frames" (since they called text regions "windows" before the GUI term was well-known)....
What I don't have, though, is a need to see all of that at once. 2-3 desktops to keep things organized (business meta-management on 1, ticketing/bug fix on 2, build consoles on 3, and VMware on 4.)
I need at most 3-4 windows visible, and rarely all of them: the active command window, for example, I only need to see a few lines around the prompt. The window(s) with a man page open, that I need to see in its entirety. Basically, I need to see all of the windows I'm reading from, but very little of the one I'm writing to. (So focus-auto-raise, a la Mac, is a truly atrocious UI feature for the way I work.)
If all the new stuff is really this restrictive, I still have the source code for Motif and my.mwmrc file from 1993.
Then we felt bad and put up a memorial to all the Chinese workers who died. It's near the Skydome, off the north-west corner. Doesn't really make it right.
Immigrant labour in the New World is, well, basically how the New World got started.
GRUB2 is certainly an upgrade for EFI-based systems. I had to download and build my own copy of GRUB2 to run Fedora from an external USB disk on a Mac. (The Bootcamp BIOS emulation thing only allows use of the first internal disk. And it means you're still dealing with BIOS.)
The GRUB2 BIOS systems I'm running seem to behave themselves, but that's only a couple of test VMs.
It's great for HTPC nodes, too. No disk on any node, all you need is a couple gig of RAM, a LAN port, and something cheap for video like an nVidia 9400M (the ION platform, basically).
Well, you can try and get people like that to make better decisions by explaining why the heavier one in blue is better.
Or you can get a company to make a better lightweight one in pink and get a cut of the action yourself. (Or you can re-skin other vendor's machines; there's a variety of businesses doing that with MacBooks; you can even get pink.)
I can pretty much guarantee the money is found by accepting people as they are and selling them something they think they want. (I don't know how to actually do that, so I remain a cog in the corporate machine.)
Switch off the theme engine in Windows 7; set it to Windows Default. Also, turn off all the little animations; there's a setting somewhere in the System control panel (I have to use the classic view to find it) where you can turn a knob to "Performance" and it kicks out a lot of the effects.
Basic rule to Windows survival: Make it look as close as you can to Windows 2000. The lower-grade your CPU or GPU is, the more important it is to do this. So it can make a huge difference on a VM, especially if you don't have 3D acceleration support.
Hmmm, just about all Linux implementations are ELF-basted, so they use a GOT (global object table), not a TOC (table of contents). AIX is probably the only "current" XCOFF-and-TOC-based OS you can get. Windows uses COFF for object files, but their load format (PE/COFF) doesn't use TOC, it uses an IAT.
So, neither Windows nor Linux have a TOC at all.
Or maybe acronyms can have more than one meaning, I always forget that....
You put the public key on the machine you're going to (in this case, the laptop).
The private key stays on the machine you run the 'ssh' command from.
And keep that private key within a single trust domain (single machine for home/school use; department LAN at the office if you must). So stealing a copy of the PUBLIC key can't compromise any other node--all you can do with it is allow someone to log on to more machines. You can't impersonate them, or log on to the machine with their private key, or any other machine with the same public key.
And, of course, you're using a passphrase on the key--though that doesn't affect the public key, only access to the private key.
I last went in a Sony store (in the Toronto Eaton Centre) about 4 years ago.
The only thing interesting about it was the miniature radio museum, showing various models of Sony radio from the iconic "transistor" set to a AM/FM/MW/LW set from the '80s.
Needless to say, they didn't have any modern equivalent for sale--they suggested the Web site.
That's easy to deal with in Postfix. Just do 'permit_sasl_authenticated' before any of the rejection rules, like 'reject_invalid_helo_hostname' or 'reject_unknown_reverse_client_hostname'.
That way, anyone who can do an authenticated SMTP session doesn't need to have rDNS or anything else set right. But MTAs, which don't use authentication, will be held to a higher standard.
I don't even bother requiring authentication on the submission port; if you meet the requirements to send through port 25, you can do it on submission as well. And vice-versa; if you can do an AUTH session on port 25, you can relay and don't have to have rDNS and so on.
(Postfix given as an example 'cause it's what I use; I would expect any decent SMTP server can do the same sort of things. If not, here's $0.00 and you can get your own copy of Postfix.)
From talking to people who actually have towers on their property, you'll do a whole lot better than just a couple of phones and a data plan. There's actual money in it for the use of your property. I know a couple in central New York State who have a hobby farm and a tower with paging and cellular on it. They've got a pretty relaxed lifestyle.
So, in a lot of places, you'll see cell antennas bolted to lower-income apartment buildings--at least in Ontario. (Ontario property tax on rental property is 2x to 3x what it is on owner-occupied, so apartments have a huge property tax burden... but are typically rented to people who can't afford to buy a condo.) The money for the cell site helps pay for building repair after the tax man has taken most of the rent away.
Then there's the "Life Free Or Fry" signs I saw in Franconia, NH....
POWER and PowerPC haven't been different since POWER3. POWER2--circa 1993--was the last "true" POWER CPU. All subsequent POWER CPUs have been based on the PowerPC ISA.
I, a system/network/development admin with integrity, say "No, you can't, here's why. Get me a signed document from Legal and I'll do it."
But they don't go to Legal. They go to someone else who says, "Yes, sure."
(Sometimes I actually deal with other people with integrity. Then I do get a document from Legal, or an, "Oops we didn't think this through, thank you." Or, one time, a transfer of copyright liability to the Lab director.)
Fortunately, most of what I do doesn't connect to Teh Real World, so I don't have to deal with people wanting to send out newsletters. I get the people trying to put GPL code into proprietary products, instead--and Legal is on my side.
Apple's not alone with 3 LAN ports; the ASUS RT-N16 has 3 LAN ports as well.
(I use the Airport Extreme as a wireless access point only; I've got the older single-band one. The RT-N16--with TomatoUSB--takes care of the 2.4 GHz clients on 802.11g/n, and the Apple does the 5.0 GHz ones. Both are stable as a rock.)
I do find that, while it isn't up to the customization of the OpenWRT, DD-WRT and Tomato firmware, the Apple firmware does a lot more than, say, a cheap Linksys will do out of the box. In particular, Apple's gadgets seem to be willing to be a router OR a bridge OR an access point; Linksys would prefer you buy three separate devices.
The Apple units are the only ones I'll run with factory firmware, though; anything else MUST run at least DD-WRT 'lite'.
Slashdot Mirror
Almost all thermal plants, regardless of heat source, use a closed steam loop. Otherwise, you have to continually de-mineralize the feedwater, and deal with precipitate in the boiler.
How the steam is condensed varies; if you're near enough water, you can just use that. If you're near enough to a town, you can sink your waste heat into the district heating system. If you're out of all other options, you can build one of those hyperboloid cooling towers people associate with nuclear plants. (They're actually used with all sorts of thermal plants, including coal and oil.)
So, even the little 550 MW combined-cycle gas-fired plant in my neighbourhood recycles steam. (Combined-cycle means the waste heat from gas turbines boils water to run a steam turbine.)
Right, and there's prior art using soundwaves as memory: mercury delay-lines.
Just don't close the loop and, instead, you have a point-to-point sound transmission system.
In the Real World, that nasty place that makes all our math hard, you'd have to deal with reflections and "impedance mismatches" (where pipes of two different sizes meet). How it would work if you went with an Aloha-net style multiple transceivers on a shared medium I'm not sure, but... I could see a decent RFC coming out of this for the day after March.
Not so much; this is more danger approaching from the side. Generally in situations where the pedestrian has right-of-way--like his example of crossing the street and someone making a left turn.
If you can hear the car coming, you can get out of the way--even when it's the car driver who should be yielding. You're still hurt even when you're not at fault.
We could ask for drivers to be held to the actual traffic law, but history is against us, here.
To further clarify the mlts' response, you hook up ONLY transmit lines on the transmitter's side. You leave out all of the handshake lines going the other way, so no RTS/CTS handshake; definitely no XON/XOFF.
If the transmitter is too fast for the receiver, the receiver will buffer-overrun and corrupt the data it sees; the UART hardware SHOULD set a status flag when it overruns. But there MUST NOT be any way for the receiver to tell the transmitter to slow down.
It is acceptable to have lines from the transmitter that tell the receiver it is "online", like hooking DTR to DSR. But it can't be done the other way; the transmitter just sends blind.
The usual way of dealing with errors in this sort of situation is to send the data multiple times. That means each message needs a sequence number. (And don't forget to include the sequence number in the checksum; so you drop the whole message if the checksum fails.)
The first step in reducing the error rate, though, is reducing the speed. RS-232 (and -422 and -423) are well-understood and quite robust--in part because they don't use really high speeds, and their data clock is much slower than their sampling clock--a nature of the asynchronous beast.
If you were really going to do this, say for modern boxes with USB, you can even get USB-to-RS-232 cables that end in pigtails or sockets and use TTL voltages; you don't need to bother with the +/-12V level conversion. Just wire RX to TX, GND to GND, and you're ready. (Bonus points for hooking DTR to DCD or something like that for detecting "remote live, no transmission".)
And if I was doing this "for real", I'd use optoisolators so the two boxes really aren't electrically connected: they don't even need to share a ground. Like the way MIDI serial works; it's an electrical current-loop like RS-232, but the receiver is an optoisolator so you don't get ground loops between your instruments and sequencer. (Once you get past that, it's basically RS-232 with an unusual clock oscillator; you can interface to it with standard UART ASICs.)
Errr. So basically what I'd do is use a communication system that is exactly like MIDI electrically (PHY, layer 1) but put my on protocol on it.
Right, where's another wheel I can reinvent?
Actually, if you pull up the same chart on Yahoo! Finance and switch it to linear scale, it looks much, much worse for Ballmer. With a linear scale, it looks like they were going to hit +INF under Gates, and it immediately plummets under Ballmer.
(And for the AC replying to you, 0.01 displayed to 1 decimal place is 0.0.)
Only problem is, underscore isn't allowed for most DNS records, especially A, AAAA and CNAME. You'll have to get anonymous-coward.xxx. (Hostnames within it are your own business, if you feel you need www.)
Him: "We're just going to start in the Cloud, we won't bother with a virtual server to start off."
*hands over server information*
Me: "Hmmm, that looks exactly like a virtual server with 'virtual server' crossed out and 'Cloud' written in with a crayon. Right down to where it uses Xen."
I've easily got 30-40 windows open at once. Console sessions on a dozen remote machines, several Web browsers with distractions and reference manuals, e-mail window, ticketing system, VMware GUI (only needed for Windows VMs, all the UNIX/Linux/BSD ones can be got to via SSH from a native terminal), a few Emacs "frames" (since they called text regions "windows" before the GUI term was well-known)....
What I don't have, though, is a need to see all of that at once. 2-3 desktops to keep things organized (business meta-management on 1, ticketing/bug fix on 2, build consoles on 3, and VMware on 4.)
I need at most 3-4 windows visible, and rarely all of them: the active command window, for example, I only need to see a few lines around the prompt. The window(s) with a man page open, that I need to see in its entirety. Basically, I need to see all of the windows I'm reading from, but very little of the one I'm writing to. (So focus-auto-raise, a la Mac, is a truly atrocious UI feature for the way I work.)
If all the new stuff is really this restrictive, I still have the source code for Motif and my .mwmrc file from 1993.
That's how we built the railroad in Canada.
Then we felt bad and put up a memorial to all the Chinese workers who died. It's near the Skydome, off the north-west corner. Doesn't really make it right.
Immigrant labour in the New World is, well, basically how the New World got started.
GRUB2 is certainly an upgrade for EFI-based systems. I had to download and build my own copy of GRUB2 to run Fedora from an external USB disk on a Mac. (The Bootcamp BIOS emulation thing only allows use of the first internal disk. And it means you're still dealing with BIOS.)
The GRUB2 BIOS systems I'm running seem to behave themselves, but that's only a couple of test VMs.
It's great for HTPC nodes, too. No disk on any node, all you need is a couple gig of RAM, a LAN port, and something cheap for video like an nVidia 9400M (the ION platform, basically).
Well, you can try and get people like that to make better decisions by explaining why the heavier one in blue is better.
Or you can get a company to make a better lightweight one in pink and get a cut of the action yourself. (Or you can re-skin other vendor's machines; there's a variety of businesses doing that with MacBooks; you can even get pink.)
I can pretty much guarantee the money is found by accepting people as they are and selling them something they think they want. (I don't know how to actually do that, so I remain a cog in the corporate machine.)
That's all right, he also claims serum cholesterol is a predictor of heart disease and doesn't back that one up at all.
Switch off the theme engine in Windows 7; set it to Windows Default. Also, turn off all the little animations; there's a setting somewhere in the System control panel (I have to use the classic view to find it) where you can turn a knob to "Performance" and it kicks out a lot of the effects.
Basic rule to Windows survival: Make it look as close as you can to Windows 2000. The lower-grade your CPU or GPU is, the more important it is to do this. So it can make a huge difference on a VM, especially if you don't have 3D acceleration support.
Hmmm, just about all Linux implementations are ELF-basted, so they use a GOT (global object table), not a TOC (table of contents). AIX is probably the only "current" XCOFF-and-TOC-based OS you can get. Windows uses COFF for object files, but their load format (PE/COFF) doesn't use TOC, it uses an IAT.
So, neither Windows nor Linux have a TOC at all.
Or maybe acronyms can have more than one meaning, I always forget that....
Uh, that's not how it works.
You put the public key on the machine you're going to (in this case, the laptop).
The private key stays on the machine you run the 'ssh' command from.
And keep that private key within a single trust domain (single machine for home/school use; department LAN at the office if you must). So stealing a copy of the PUBLIC key can't compromise any other node--all you can do with it is allow someone to log on to more machines. You can't impersonate them, or log on to the machine with their private key, or any other machine with the same public key.
And, of course, you're using a passphrase on the key--though that doesn't affect the public key, only access to the private key.
I last went in a Sony store (in the Toronto Eaton Centre) about 4 years ago.
The only thing interesting about it was the miniature radio museum, showing various models of Sony radio from the iconic "transistor" set to a AM/FM/MW/LW set from the '80s.
Needless to say, they didn't have any modern equivalent for sale--they suggested the Web site.
And it can fall back to SMS/MMS or e-mail--you can pick which in settings.
...but... I don't know who to trust.... No, that's wrong, I trust no-one.
That's easy to deal with in Postfix. Just do 'permit_sasl_authenticated' before any of the rejection rules, like 'reject_invalid_helo_hostname' or 'reject_unknown_reverse_client_hostname'.
That way, anyone who can do an authenticated SMTP session doesn't need to have rDNS or anything else set right. But MTAs, which don't use authentication, will be held to a higher standard.
I don't even bother requiring authentication on the submission port; if you meet the requirements to send through port 25, you can do it on submission as well. And vice-versa; if you can do an AUTH session on port 25, you can relay and don't have to have rDNS and so on.
(Postfix given as an example 'cause it's what I use; I would expect any decent SMTP server can do the same sort of things. If not, here's $0.00 and you can get your own copy of Postfix.)
From talking to people who actually have towers on their property, you'll do a whole lot better than just a couple of phones and a data plan. There's actual money in it for the use of your property. I know a couple in central New York State who have a hobby farm and a tower with paging and cellular on it. They've got a pretty relaxed lifestyle.
So, in a lot of places, you'll see cell antennas bolted to lower-income apartment buildings--at least in Ontario. (Ontario property tax on rental property is 2x to 3x what it is on owner-occupied, so apartments have a huge property tax burden... but are typically rented to people who can't afford to buy a condo.) The money for the cell site helps pay for building repair after the tax man has taken most of the rent away.
Then there's the "Life Free Or Fry" signs I saw in Franconia, NH....
POWER and PowerPC haven't been different since POWER3. POWER2--circa 1993--was the last "true" POWER CPU. All subsequent POWER CPUs have been based on the PowerPC ISA.
http://en.wikipedia.org/wiki/IBM_POWER#POWER3
Here's how it works, because I've seen it happen:
I, a system/network/development admin with integrity, say "No, you can't, here's why. Get me a signed document from Legal and I'll do it."
But they don't go to Legal. They go to someone else who says, "Yes, sure."
(Sometimes I actually deal with other people with integrity. Then I do get a document from Legal, or an, "Oops we didn't think this through, thank you." Or, one time, a transfer of copyright liability to the Lab director.)
Fortunately, most of what I do doesn't connect to Teh Real World, so I don't have to deal with people wanting to send out newsletters. I get the people trying to put GPL code into proprietary products, instead--and Legal is on my side.
Whoops, I'm an idiot today; the RT-N16 has three _antennas_; 4 LAN ports, just like the GP said is typical.
(My first idiocy was ordering a motorcycle tire based on the Honda parts catalogue, not based on what size rim is actually there....)
Apple's not alone with 3 LAN ports; the ASUS RT-N16 has 3 LAN ports as well.
(I use the Airport Extreme as a wireless access point only; I've got the older single-band one. The RT-N16--with TomatoUSB--takes care of the 2.4 GHz clients on 802.11g/n, and the Apple does the 5.0 GHz ones. Both are stable as a rock.)
I do find that, while it isn't up to the customization of the OpenWRT, DD-WRT and Tomato firmware, the Apple firmware does a lot more than, say, a cheap Linksys will do out of the box. In particular, Apple's gadgets seem to be willing to be a router OR a bridge OR an access point; Linksys would prefer you buy three separate devices.
The Apple units are the only ones I'll run with factory firmware, though; anything else MUST run at least DD-WRT 'lite'.