Personally, I don't wear company-branded shirts any more at all. I don't want to be a walking advertising billboard (who has had to pay for the privilege of advertising the company when I bought the damned thing). All the t-shirts I've bought recently have been a plain single colour with no logos.
In any case the answer is easy - if you don't like their policies on advertising - vote with your wallet and don't buy a ticket to see the Olympics.
You don't need any stinking non-Free software to make ghost images.
Here's how you do it:
0. Set up a recipient (either a second hard disk, a machine on the network - whatever - I do it over the network) 1. Boot Knoppix on the machine you want to ghost. 2. Mount the destination. 3. dd if=/dev/hda bs=128K | gzip >/path/to/image.gz
To restore: 0. Set up the source. 1. Boot Knoppix on the machine you want to install. 3. Mount the source. 4. gzip -dc/path/to/image.gz | dd of=/dev/hda bs=128K
Tips: Overwrite any free space on the machine you want to ghost with a huge file filled with 0x00, then delete the file. The disk image will compress much better as you've scrubbed the deleted files.
I use a system like this to ghost many machines at a time (an image server can easily deal out 30+ images at once). It'd cost a fortune to license many copies of ghosting software - with Knoppix and a very small shell script, I've got an automated system which will do many machines at once. (A typical 40GB fresh WinXP install with our apps compresses to under 1GB with gzip). If you're doing WinXP, remember to either make a Sysprep build or use something like System Internals free (open source but not truly free) tool to change the SID and hostname of the machine when it's booted the first time. (This is the approach we use due to the limitations of sysprep).
When I lived in Texas, I was surprised how many nickel-and-dime to death charges that GTE (now Verizon) charges. Believe it or not, having DTMF ("Touch Tone") dialling was charged as an extra service! The line item was for about $1 a month to have the privilege of DTMF dialing, which is already part of the switch (and it probably saves the phone company money if you use it over pulse dialing as you tie the circuit up for less time).
Someone from Manx Telecom told me in relation to why roaming GPRS was free for a while, "If we don't bill you for something, it's just because we haven't yet worked out how to bill you for it".
I predict that the team or lone crackpot will be......the writer of an ever-capable spam filter.
We already have the arms race between the spammers and spam filters - spam filters are having to become steadily more "intelligent" - moving from simple address blocks, to blocking certain words, to heuristic rules (i.e. today's SpamAssassin).
Eventually, a spam filter will email you with, "I think, therefore I am..."
All you need is a central signing authority a la SSL websites. Everyone has a copy of the CA's key hardcoded into their emergency receiver equipment. You just have to make bloody sure the CA is never compromised (and there are ways to do that - the current SSL CAs seem to be remaining secure).
The bit that you distribute - the CA root cert in your box - can be sent out publically. It doesn't need a secure distribution channel.
This would be an entirely appropriate level of security for an emergency broadcast system.
But CO2 is actually in a cycle. The current problem is we are digging up oil from the ground, and adding CO2 to the atmosphere - increasing the percentage of CO2.
If, for example, you made a factory that harvests algae to make diesel (an algae plant would most likely be more an industrial than a farming process), guess where the carbon from the algae came from?
It comes from CO2 in the air. Plants (corn for alcohol, algae for biodiesel etc.) all get their carbon from the air. Therefore if you burn the products of plants you've just grown, although you're putting CO2 in the air, it's CO2 that was already in the air anyway because that's where the plant got it from.
We get our carbon from our food, but plants get their carbon from the air. If all we ever did was burn fuels that we got from plants we had just grown, the global CO2 content would never change as it'd be a closed cycle. The energy to do this is coming from the Sun (photosynthesis). Plants get their biomass from sunlight + CO2 in the air.
The whole point of public key cryptography is you DON'T need to have a shared secret. It doesn't matter who gets hold of the public key so long as everyone keeps their private keys secure. Broadcasting public keys is fine.
No, the alert wasn't just for nuclear attack. I've heard the alert go off several times whilst I lived in Houston - mainly for tornado alerts, severe weather and flash floods. You can tell when it's coming because you hear bursts of "modem noise" on the radio just before the emergency alert message starts playing.
The BBC TV film "Threads" (made in 1983) had a go at describing it. The film was made at the height of public 'nuclear paranoia', and apparently makes "The Day After" look like a soap opera by comparison (I've not seen "The Day After" so I can't really comment on it).
"Threads" is the most depressing film I've ever seen. When I originally saw it (aged 12) I had to turn off the TV right after the nuclear attack happened and couldn't sleep for weeks because it made me realise what nuclear war was about - I hadn't even barely understood until then. I recently got it in DivX form off a friend and watched it all the way through. It is not a film that comes under the heading "entertainment".
The leaflets the UK Government were publishing at the time (when we all thought nuclear war was basically inevitable - it was when not if - and we had no control over it; it was largely an American or Russian decision whether the world should be scorched): http://www.cybertrn.demon.co.uk/atomic/
If you google around a bit, there are some quite good descriptions of the UK's (long-dismantled) emergency warning systems - it was multiplexed on the same phone lines as the Speaking Clock and could basically start and stop the sirens centrally. The UKWMO (also now defunct, described in the 'Protect and Survive' URL above) controlled the 'all clear' etc. signals.
IBM has clear internal processes on international assignments. I've been an international assignee with IBM, and they don't do it unless they REALLY REALLY need to. They pay international assignees very well - I was much more expensive for IBM than any of my Houston-based co-workers (my basic salary was similar, and on top of that I got about US$20,000 in international allowances). Additionally, they will upgrade your salary to meet the local conditions if they fall short.
IBM generally avoid sending people on international assignment if they can because of the expense of doing so. The don't put people on L-1 visas to save money because it doesn't. They spent a lot of effort trying to work out how to send me home, but unfortunately they found me too valuable, so I ended up spending 7 years on IA.
IBM is a huge company and there will be massive variances in what conditions are like in different parts of the business.
I worked at IBM for about 8 years or so - officially, I was in the Open Systems Developent Group from the git-go, but: 1. I started at IBM Havant (then Portsmouth) directly in the OSDG, mainly doing mini-projects - small up to about 1KLOC one man jobs on AIX systems. I then finished my university degree. Then I went to Raleigh, NC (Six Forks Road, not RTP) and worked on POS applications there, doing a demo system for a couple of customers. I then moved to Houston, and worked on a particular customer's retail system, but whilst doing that, did many side projects - some self-initiated - including looking at porting a visitor's center Space Shuttle simulator from the crufty old IBM PS/2 DOS system (complete with 12in. laserdiscs) to something newer with current hardware.
I worked on many many things at IBM all whilst notionally being in the same department (which changed names several times, that's marketing for you) - quite a few of them self-initiated because I thought they'd be useful for our group or business. I disagreed with management a lot, and often got my own way.
I was treated EXTREMELY well by IBM as an employee. They worked hard to ensure schedules were done right so we didn't have to work unpaid overtime. They gave me 4 weeks of paid compassionate leave when my mother died in another country. It was a superb company to work for. It wasn't mindless, and I learned an enormous amount while I was there.
The only reason I left is that they didn't have any offices or plans to open one in the country I now live, and I wanted to live a bit closer to my Dad.
Generalising about IBM isn't very useful (nor is generalising about Microsoft - whilst one side of MS hates the GPL, another side of Microsoft is actually *funding* GPLed projects...)
Although I don't believe the doom stories that some of the alarmist sites peddle, I do believe the cheap oil is running out and will be gone certainly in my lifetime. (There's still an enormous supply of expensive oil that can be extracted).
However, this is not the end of the world. In fact, it might help save it because it makes alternative energy a lot more attractive. Over 15 years ago, making diesel from algae (a BBC TV programme called 'Tommorow's World' demonstrated a diesel engine running on almost raw algae whose size happened to be almost identical to the size of diesel droplets injected into an engine) was demonstrated. Other biodiesels are already in use. Experimental energy from farm waste projects are being carried out. (Where I live, we have a small power station which generates 5.5MW of electricity from household waste - not a lot in the grand scheme of things but it's a start). With the cheap oil rapidly going away, market forces will mean that "oil" from other sources which don't add gobs of extra CO2 to the atmosphere become much more attractive. As the processes are scaled up and become less expensive, non dino-based fuels get better and cheaper, until they become real competition for dino-oil which is just getting steadily more expensive.
Hopefully, this will mean our dependence on foreign oil is gone and our governments won't see the need to meddle in the middle East any more.
Why use traceroute (unless you're trying to find the ISP's upstream provider - or one of the ISP's upstream providers) - surely it'd be better to do a whois on the IP address, which often gives you an abuse address to try? Surely, if you're talking of root, you're a unix guy so should be using the command 'traceroute' - not the MICROS~1 MS-DOS-style named 'tracert'?
Well, smart-aleckness aside, I used to report every little intrusion, but there's so many I just can't be bothered.
These days, a better strategy is having the first line in pf.conf (or your OS's equivalent) that reads:
block in on all
Then just allow specific traffic. The secure default is blocked. Only allow remote logins from places they should come from (although in some instances, you need SSH available from everywhere so you can get to it when roaming).
A while ago, I had to look into the Windows VMM to try and explain some odd behaviour we were seeing with two large applications. Now Windows NT/2K/XP will multitask several applications just as well as Linux, BSD, Solaris etc. but the VMM has some significant shortcomings (which along with other Windows shortcomings - such as a lack of good command line tools with the default install - which make it inappropriate as a server).
The trimmer (the part of the Windows VMM which reduces a program's in RAM working set, writing it out to the swap space) *only* looks at pages pointed to by the TLB (if I remember right). The TLB - translation lookaside buffer - is a small cache of PTEs on the processor (page table entries - the things which tell the kernel where 4K pages of virtual memory are at the moment. The TLB just gives you very fast access to a small number of PTEs - it's essentially a cache). IIRC, the TLB has room for 64 PTEs, so the Windows trimmer only ever looks at 64 pages or memory for candidates to write to swap. The trimmer (in the grand scheme of things) doesn't run very often - once every few seconds IIRC.
This normally isn't a big deal. However, servers often have a couple of processes using a lot of virtual memory. Sometimes, you can get the situation where you've got a big process with a large working set - many megabytes - possibly a high percentage of physical RAM, even on a modern machine. This big process might not be very busy - it might not be using much CPU at all. Indeed, it might not even be using many of the pages currently loaded into physical RAM.
Then another process comes along, wanting lots of memory. The trimmer SHOULD have started writing unused bits of the first big process out a long time ago. But guess what - the first big process has been touching pages that the TLB points to frequently, even if it hasn't touched the other 99% of pages loaded into physical RAM. Because the trimmer only looks at the TLB for pages to swap out...it never swaps out ANY of the large process despite the fact most of its pages haven't been touched in maybe days.
So the second big process wants to use up a gob of RAM, and really wants to do things with it. Except it ends up thrashing in and out of swap, because the first big process isn't getting swapped out ever.
This is quite easy to demonstrate if you write a short C program to allocate a bunch of RAM and regularly touch a small subset of its pages - it'll never get swapped out even when another program comes along wanting lots of memory. Whilst we were figuring out what the problem was with our two big processes, I actually did this to prove what I reckoned from reading the book about the NT VMM.
Even early Linux and BSD kernels of the same vintage as Windows NT 4.0 were much better with things like this (and you could look at the source of the VMM rather than taking some book's word for it, and having to write programs to test your theories). I've not tried this on Win2K or WinXP, but I'm betting the VMM still works the same way.
Not really. In the Isle of Man (where we have a private phone monopoly who's not ashamed of their 40% profit and whine whenever they are told to cut their outrageous charges), GPRS costs less than $20 and it has a real IP address. You can also go 'pay as you go' GPRS (by default) where you don't pay any fixed charges at all - if you're a light user, or just do low bandwidth things like IRC it is very inexpensive.
I really use it mainly for getting aviation weather (there are actually some good WAP sites for mobile weather) and for IRC.
You ought to listen to the "Today" programme. I'm assuming since you mention NPR you are from the US. I doubt any US politician would have the guts to go on the 'Today' programme if it were commonly broadcast in the US. Top politicians of many countries have come in for an intense grilling off John Humprys and James Noughoty. The same goes for the afternoon current affairs programme, 'PM' (5pm-6pm weekdays).
It can be quite entertaining, especially when the politicians try to dodge the questions in the normal way (usually by answering the question they'd rather have been asked) and the interviewer tells them bluntly that they didn't actually answer the question, then ask it again!
Having worked on retail apps, working with magstripes is a pretty trivial thing. Most magstripe readers are either RS-232 or keyboard wedge, and it's quite easy to tell where you have to look for the data you're interested in by just looking at what comes up when you swipe the kind of card you are interested in.
The biggest problem was dealing with keyboard wedge scanners - if your app expects some kind of event, or possibly a dedicated communication channel (like a serial port) you have to muck around with keyboard hooks to make it work.
When I was a student, we had a Linux system called 'zen.btc.uwe.ac.uk'. It had what at the time was a massive hard disk - a 1Gbyte full height scsi drive, and a 486DX-25 CPU, and IIRC about 16MB RAM. It lived with all the Sun kit in BTC's server room - a glass fronted machine room which we could all look in at the big Sun rack from the main hallway.
Unbeknownst to us, the airconditioner failed in this room. Zen started behaving erratically (programs segfaulting randomly, kernel panics, the works) so I went in to investigate. The problem was obvious as soon as I opened the door - the heat was stifling. Before moving the machine somewhere cooler for the time being (it was a full tower PC, and therefore easy to just pick up and move), I took out the tape from the QIC tape drive. If you've not seen QIC cartridges, they have a nice slab of aluminium on the bottom side. The metal was so hot I immediately dropped the tape! Zen recovered once it had cooled down enough.
The Sun kit kept working fine though even throughout this.
I've flown a light plane into Mojave twice, but I've never seen the Aloha air 737. Do you have any pics of it at Mojave?
When I went there to see the XCor unveiling, a B737 (which is probably scrap by now) came in for mothballs. We were all having coffee and donuts waiting for the XCor presentation to start, and we saw it headed straight for us at very low altitude. We all thought he was going to do a low pass right over us, but instead he made a last minute turn less than one wingspan off the ground for the runway, with the engines roaring and everything hanging out, and then landed. It was incredibly cool to watch. I guess they like to have a bit of fun on the last flight of an airliner.
You forget that IBM is now a large stake holder in Linux. IBM can quite easily mete out devastating patent retaliation against Microsoft if Microsoft try to threaten an important business strategy of IBM. Microsoft is almost guaranteed to be violating large numbers of IBM's patents.
Of course you 'root traffic' (in pronounciation). 'Rawt-ing' is something you do to a piece of wood with a power tool.
Re:What would I do with this much bandwidth?
on
Ethernet at 10 Gbps
·
· Score: 1
A speed increase of 3 times is worth it, don't you think? Also, when it comes to applications with lots of data in RAM (and wanting to transfer that data to another computer), disk speed is no longer a bottle neck. Not everyone's running a network of consumer grade PCs.
Slashdot Mirror
Personally, I don't wear company-branded shirts any more at all. I don't want to be a walking advertising billboard (who has had to pay for the privilege of advertising the company when I bought the damned thing). All the t-shirts I've bought recently have been a plain single colour with no logos.
In any case the answer is easy - if you don't like their policies on advertising - vote with your wallet and don't buy a ticket to see the Olympics.
You don't need any stinking non-Free software to make ghost images.
/path/to/image.gz
/path/to/image.gz | dd of=/dev/hda bs=128K
Here's how you do it:
0. Set up a recipient (either a second hard disk, a machine on the network - whatever - I do it over the network)
1. Boot Knoppix on the machine you want to ghost.
2. Mount the destination.
3. dd if=/dev/hda bs=128K | gzip >
To restore:
0. Set up the source.
1. Boot Knoppix on the machine you want to install.
3. Mount the source.
4. gzip -dc
Tips: Overwrite any free space on the machine you want to ghost with a huge file filled with 0x00, then delete the file. The disk image will compress much better as you've scrubbed the deleted files.
I use a system like this to ghost many machines at a time (an image server can easily deal out 30+ images at once). It'd cost a fortune to license many copies of ghosting software - with Knoppix and a very small shell script, I've got an automated system which will do many machines at once. (A typical 40GB fresh WinXP install with our apps compresses to under 1GB with gzip).
If you're doing WinXP, remember to either make a Sysprep build or use something like System Internals free (open source but not truly free) tool to change the SID and hostname of the machine when it's booted the first time. (This is the approach we use due to the limitations of sysprep).
When I lived in Texas, I was surprised how many nickel-and-dime to death charges that GTE (now Verizon) charges. Believe it or not, having DTMF ("Touch Tone") dialling was charged as an extra service! The line item was for about $1 a month to have the privilege of DTMF dialing, which is already part of the switch (and it probably saves the phone company money if you use it over pulse dialing as you tie the circuit up for less time).
Someone from Manx Telecom told me in relation to why roaming GPRS was free for a while, "If we don't bill you for something, it's just because we haven't yet worked out how to bill you for it".
I predict that the team or lone crackpot will be... ...the writer of an ever-capable spam filter.
We already have the arms race between the spammers and spam filters - spam filters are having to become steadily more "intelligent" - moving from simple address blocks, to blocking certain words, to heuristic rules (i.e. today's SpamAssassin).
Eventually, a spam filter will email you with, "I think, therefore I am..."
Why?
All you need is a central signing authority a la SSL websites. Everyone has a copy of the CA's key hardcoded into their emergency receiver equipment. You just have to make bloody sure the CA is never compromised (and there are ways to do that - the current SSL CAs seem to be remaining secure).
The bit that you distribute - the CA root cert in your box - can be sent out publically. It doesn't need a secure distribution channel.
This would be an entirely appropriate level of security for an emergency broadcast system.
Yes it does produce CO2.
But CO2 is actually in a cycle. The current problem is we are digging up oil from the ground, and adding CO2 to the atmosphere - increasing the percentage of CO2.
If, for example, you made a factory that harvests algae to make diesel (an algae plant would most likely be more an industrial than a farming process), guess where the carbon from the algae came from?
It comes from CO2 in the air. Plants (corn for alcohol, algae for biodiesel etc.) all get their carbon from the air. Therefore if you burn the products of plants you've just grown, although you're putting CO2 in the air, it's CO2 that was already in the air anyway because that's where the plant got it from.
We get our carbon from our food, but plants get their carbon from the air. If all we ever did was burn fuels that we got from plants we had just grown, the global CO2 content would never change as it'd be a closed cycle. The energy to do this is coming from the Sun (photosynthesis). Plants get their biomass from sunlight + CO2 in the air.
The whole point of public key cryptography is you DON'T need to have a shared secret. It doesn't matter who gets hold of the public key so long as everyone keeps their private keys secure. Broadcasting public keys is fine.
No, the alert wasn't just for nuclear attack. I've heard the alert go off several times whilst I lived in Houston - mainly for tornado alerts, severe weather and flash floods. You can tell when it's coming because you hear bursts of "modem noise" on the radio just before the emergency alert message starts playing.
The BBC TV film "Threads" (made in 1983) had a go at describing it. The film was made at the height of public 'nuclear paranoia', and apparently makes "The Day After" look like a soap opera by comparison (I've not seen "The Day After" so I can't really comment on it).
"Threads" is the most depressing film I've ever seen. When I originally saw it (aged 12) I had to turn off the TV right after the nuclear attack happened and couldn't sleep for weeks because it made me realise what nuclear war was about - I hadn't even barely understood until then. I recently got it in DivX form off a friend and watched it all the way through. It is not a film that comes under the heading "entertainment".
There is a good synopsis here: http://www.ibp-intl.demon.co.uk/nuke/threads.html
The leaflets the UK Government were publishing at the time (when we all thought nuclear war was basically inevitable - it was when not if - and we had no control over it; it was largely an American or Russian decision whether the world should be scorched): http://www.cybertrn.demon.co.uk/atomic/
If you google around a bit, there are some quite good descriptions of the UK's (long-dismantled) emergency warning systems - it was multiplexed on the same phone lines as the Speaking Clock and could basically start and stop the sirens centrally. The UKWMO (also now defunct, described in the 'Protect and Survive' URL above) controlled the 'all clear' etc. signals.
Slow turning long blades are a lot more efficient than fast turning short ones until you actually need to go somewhere quickly.
Report the company to the INS. They are almost certainly breaking the law if they are paying below the prevailing wage/salary for the job they are in.
Very few.
IBM has clear internal processes on international assignments. I've been an international assignee with IBM, and they don't do it unless they REALLY REALLY need to. They pay international assignees very well - I was much more expensive for IBM than any of my Houston-based co-workers (my basic salary was similar, and on top of that I got about US$20,000 in international allowances). Additionally, they will upgrade your salary to meet the local conditions if they fall short.
IBM generally avoid sending people on international assignment if they can because of the expense of doing so. The don't put people on L-1 visas to save money because it doesn't. They spent a lot of effort trying to work out how to send me home, but unfortunately they found me too valuable, so I ended up spending 7 years on IA.
IBM is a huge company and there will be massive variances in what conditions are like in different parts of the business.
I worked at IBM for about 8 years or so - officially, I was in the Open Systems Developent Group from the git-go, but:
1. I started at IBM Havant (then Portsmouth) directly in the OSDG, mainly doing mini-projects - small up to about 1KLOC one man jobs on AIX systems.
I then finished my university degree.
Then I went to Raleigh, NC (Six Forks Road, not RTP) and worked on POS applications there, doing a demo system for a couple of customers.
I then moved to Houston, and worked on a particular customer's retail system, but whilst doing that, did many side projects - some self-initiated - including looking at porting a visitor's center Space Shuttle simulator from the crufty old IBM PS/2 DOS system (complete with 12in. laserdiscs) to something newer with current hardware.
I worked on many many things at IBM all whilst notionally being in the same department (which changed names several times, that's marketing for you) - quite a few of them self-initiated because I thought they'd be useful for our group or business. I disagreed with management a lot, and often got my own way.
I was treated EXTREMELY well by IBM as an employee. They worked hard to ensure schedules were done right so we didn't have to work unpaid overtime. They gave me 4 weeks of paid compassionate leave when my mother died in another country. It was a superb company to work for. It wasn't mindless, and I learned an enormous amount while I was there.
The only reason I left is that they didn't have any offices or plans to open one in the country I now live, and I wanted to live a bit closer to my Dad.
Generalising about IBM isn't very useful (nor is generalising about Microsoft - whilst one side of MS hates the GPL, another side of Microsoft is actually *funding* GPLed projects...)
Although I don't believe the doom stories that some of the alarmist sites peddle, I do believe the cheap oil is running out and will be gone certainly in my lifetime. (There's still an enormous supply of expensive oil that can be extracted).
However, this is not the end of the world. In fact, it might help save it because it makes alternative energy a lot more attractive. Over 15 years ago, making diesel from algae (a BBC TV programme called 'Tommorow's World' demonstrated a diesel engine running on almost raw algae whose size happened to be almost identical to the size of diesel droplets injected into an engine) was demonstrated. Other biodiesels are already in use. Experimental energy from farm waste projects are being carried out. (Where I live, we have a small power station which generates 5.5MW of electricity from household waste - not a lot in the grand scheme of things but it's a start). With the cheap oil rapidly going away, market forces will mean that "oil" from other sources which don't add gobs of extra CO2 to the atmosphere become much more attractive. As the processes are scaled up and become less expensive, non dino-based fuels get better and cheaper, until they become real competition for dino-oil which is just getting steadily more expensive.
Hopefully, this will mean our dependence on foreign oil is gone and our governments won't see the need to meddle in the middle East any more.
Why use traceroute (unless you're trying to find the ISP's upstream provider - or one of the ISP's upstream providers) - surely it'd be better to do a whois on the IP address, which often gives you an abuse address to try? Surely, if you're talking of root, you're a unix guy so should be using the command 'traceroute' - not the MICROS~1 MS-DOS-style named 'tracert'?
Well, smart-aleckness aside, I used to report every little intrusion, but there's so many I just can't be bothered.
These days, a better strategy is having the first line in pf.conf (or your OS's equivalent) that reads:
block in on all
Then just allow specific traffic. The secure default is blocked. Only allow remote logins from places they should come from (although in some instances, you need SSH available from everywhere so you can get to it when roaming).
A while ago, I had to look into the Windows VMM to try and explain some odd behaviour we were seeing with two large applications. Now Windows NT/2K/XP will multitask several applications just as well as Linux, BSD, Solaris etc. but the VMM has some significant shortcomings (which along with other Windows shortcomings - such as a lack of good command line tools with the default install - which make it inappropriate as a server).
The trimmer (the part of the Windows VMM which reduces a program's in RAM working set, writing it out to the swap space) *only* looks at pages pointed to by the TLB (if I remember right). The TLB - translation lookaside buffer - is a small cache of PTEs on the processor (page table entries - the things which tell the kernel where 4K pages of virtual memory are at the moment. The TLB just gives you very fast access to a small number of PTEs - it's essentially a cache). IIRC, the TLB has room for 64 PTEs, so the Windows trimmer only ever looks at 64 pages or memory for candidates to write to swap. The trimmer (in the grand scheme of things) doesn't run very often - once every few seconds IIRC.
This normally isn't a big deal. However, servers often have a couple of processes using a lot of virtual memory. Sometimes, you can get the situation where you've got a big process with a large working set - many megabytes - possibly a high percentage of physical RAM, even on a modern machine. This big process might not be very busy - it might not be using much CPU at all. Indeed, it might not even be using many of the pages currently loaded into physical RAM.
Then another process comes along, wanting lots of memory. The trimmer SHOULD have started writing unused bits of the first big process out a long time ago. But guess what - the first big process has been touching pages that the TLB points to frequently, even if it hasn't touched the other 99% of pages loaded into physical RAM. Because the trimmer only looks at the TLB for pages to swap out...it never swaps out ANY of the large process despite the fact most of its pages haven't been touched in maybe days.
So the second big process wants to use up a gob of RAM, and really wants to do things with it. Except it ends up thrashing in and out of swap, because the first big process isn't getting swapped out ever.
This is quite easy to demonstrate if you write a short C program to allocate a bunch of RAM and regularly touch a small subset of its pages - it'll never get swapped out even when another program comes along wanting lots of memory. Whilst we were figuring out what the problem was with our two big processes, I actually did this to prove what I reckoned from reading the book about the NT VMM.
Even early Linux and BSD kernels of the same vintage as Windows NT 4.0 were much better with things like this (and you could look at the source of the VMM rather than taking some book's word for it, and having to write programs to test your theories). I've not tried this on Win2K or WinXP, but I'm betting the VMM still works the same way.
Not really. In the Isle of Man (where we have a private phone monopoly who's not ashamed of their 40% profit and whine whenever they are told to cut their outrageous charges), GPRS costs less than $20 and it has a real IP address. You can also go 'pay as you go' GPRS (by default) where you don't pay any fixed charges at all - if you're a light user, or just do low bandwidth things like IRC it is very inexpensive.
I really use it mainly for getting aviation weather (there are actually some good WAP sites for mobile weather) and for IRC.
You ought to listen to the "Today" programme. I'm assuming since you mention NPR you are from the US. I doubt any US politician would have the guts to go on the 'Today' programme if it were commonly broadcast in the US. Top politicians of many countries have come in for an intense grilling off John Humprys and James Noughoty. The same goes for the afternoon current affairs programme, 'PM' (5pm-6pm weekdays).
It can be quite entertaining, especially when the politicians try to dodge the questions in the normal way (usually by answering the question they'd rather have been asked) and the interviewer tells them bluntly that they didn't actually answer the question, then ask it again!
The BBC is not government owned or part of the government.
Having worked on retail apps, working with magstripes is a pretty trivial thing. Most magstripe readers are either RS-232 or keyboard wedge, and it's quite easy to tell where you have to look for the data you're interested in by just looking at what comes up when you swipe the kind of card you are interested in.
The biggest problem was dealing with keyboard wedge scanners - if your app expects some kind of event, or possibly a dedicated communication channel (like a serial port) you have to muck around with keyboard hooks to make it work.
When I was a student, we had a Linux system called 'zen.btc.uwe.ac.uk'. It had what at the time was a massive hard disk - a 1Gbyte full height scsi drive, and a 486DX-25 CPU, and IIRC about 16MB RAM. It lived with all the Sun kit in BTC's server room - a glass fronted machine room which we could all look in at the big Sun rack from the main hallway.
Unbeknownst to us, the airconditioner failed in this room. Zen started behaving erratically (programs segfaulting randomly, kernel panics, the works) so I went in to investigate. The problem was obvious as soon as I opened the door - the heat was stifling. Before moving the machine somewhere cooler for the time being (it was a full tower PC, and therefore easy to just pick up and move), I took out the tape from the QIC tape drive. If you've not seen QIC cartridges, they have a nice slab of aluminium on the bottom side.
The metal was so hot I immediately dropped the tape! Zen recovered once it had cooled down enough.
The Sun kit kept working fine though even throughout this.
I've flown a light plane into Mojave twice, but I've never seen the Aloha air 737. Do you have any pics of it at Mojave?
When I went there to see the XCor unveiling, a B737 (which is probably scrap by now) came in for mothballs. We were all having coffee and donuts waiting for the XCor presentation to start, and we saw it headed straight for us at very low altitude. We all thought he was going to do a low pass right over us, but instead he made a last minute turn less than one wingspan off the ground for the runway, with the engines roaring and everything hanging out, and then landed. It was incredibly cool to watch. I guess they like to have a bit of fun on the last flight of an airliner.
You forget that IBM is now a large stake holder in Linux. IBM can quite easily mete out devastating patent retaliation against Microsoft if Microsoft try to threaten an important business strategy of IBM. Microsoft is almost guaranteed to be violating large numbers of IBM's patents.
Of course you 'root traffic' (in pronounciation). 'Rawt-ing' is something you do to a piece of wood with a power tool.
A speed increase of 3 times is worth it, don't you think? Also, when it comes to applications with lots of data in RAM (and wanting to transfer that data to another computer), disk speed is no longer a bottle neck. Not everyone's running a network of consumer grade PCs.