In this case the facts are that the software files in question where under a BSD license Wrong. It was dual-licensed code with an "alternatively" stuck between them.
That also means that attempting to add a GPL or other license isn't permitted. Wrong. The list is not an exhaustive list (it doesn't say you must follow those and only those conditions). Any license that is compatible may be added, as the BSD says nothing about additional restrictions.
However, that is not the full license text in question. It was a dual-license that also said:
Alternatively, this software may be distributed under the terms of the GNU General Public License ("GPL") version 2 as published by the Free Software Foundation.
Most people belive that means you can either accept the BSD terms or accept the GPL terms (and from then on follow only the one chosen set of terms). Theo seems to be claiming that you somehow have to follow both sets of terms. I guess it depends on your definition of "alternatively".
If you released something under the BSD license and someone made a closed-source commercial program out of it (as allowed under the BSD license and done many times by many companies), would you be incensed? If so, why would you release something under a license that allows others to do something you don't want?
The current 3 clause BSD license allows someone to release derived works under the GPL (or under closed-source commercial license). If you don't like that, then don't use the 3 clause BSD license. Licenses have specific meaning that should be understood before they are used.
Yeah, I typed that while doing something else (working on a new Nagios config) and added an extra zero.
Still, the point stands. Residential customers are not going to pay for dedicated (non-oversold) bandwidth. The costs involved in attempting to provide dedicated bandwidth for everybody would cause the prices to go up as massive amounts of equipment is replaced.
If you don't want oversold bandwidth, pull out your checkbook. The ISP I work for would be happy to sell you a 30 megabits/sec link, but it'll probably cost you around $60000/month. If you want us to "upgrade the infrastructure" to handle selling 30 megabits/sec to thousands of customers, somebody's going to have to pay that bill as well (just off the top of my head, I'd say it'll probably at least triple the cost).
We oversell DSL, but we monitor the links to make sure we have sufficient capacity for all but the most extreme peaks (the typical average bits/customer for our DSL is only around 35-40 kilobits/sec, a small fraction of each customer's available bandwidth). This is the way pretty much all modern communication networks work; land-line and cell phone networks cannot handle everybody picking up the phone and making a call at once.
The uninformed user knows Cisco as "the network company that the Internet is connected with." Yeah, but the informed user knows Cisco sucks and would rather have a Juniper.:-)
Had someone told me 10 years ago that I could own *my very own* full-featured Cisco router for under $100, I would've given a finger to sign up. Changing the name and artwork on the box doesn't magically make the Linksys routers full-featured Cisco routers. What most people consider full-featured Cisco includes IOS, and I doubt they're planning to put IOS on a home wireless router (certainly not for $50).
At this point they'd be too far behind; it takes years to bring a new chip design to market. The current AMD and Intel chips would beat any Alpha in price/performance. I wish I could get x86 systems with the design effort of the Alpha systems though; it wasn't all about the CPU.
The Alpha and VMS didn't die with DEC (or, rather, Digital). When HP bought them, they realised that having two CPU families developed in-house (Alpha and PA-RISC) was not economically viable, and partnered with Intel to replace them both with Itanium. You've got the chronology wrong. HP partnered with Intel on Itanium (planned to replace PA-RISC and x86 both) while DEC sold Alphas. DEC was bought by Compaq, who then decided to end Alpha development after the EV7 generation and move to Itanium for both VMS and Tru64 Unix. Then HP bought Compaq, continued the VMS move to Itanium, committed to merging Tru64 features with HP-UX on Itanium, then dropped that and said Tru64 users should just move to HP-UX on Itanium (yeah right, I'm going to give them more money). They stopped selling Alpha systems a few months ago.
Not quite the same, but just last week I was over at my parents' house. We noticed the lights flickering (and the UPSes clicking) every 30-90 seconds or so for about half an hour, so we stuck a voltmeter in the socket. It was at 125V; when the lights flickered, it dropped to 120V for 30 seconds, the lights flickered again, and it went back to 125V. It did this for an hour or two before the power finally went out for an hour. The utility company had some type of failure at the substation that didn't die right away.
Yeah, but then when you get three threads, it gets more complicated. Also, one indecisive thread, and everybody stands around and complains while the one tries to decide which slice to take.
Well, the same can be said about a solar power plant, but that wasn't really my point. To replace the nuclear plant with a similar solar plant, you would have to have a solar plant that covered 125 square miles (based on the size and power listed in the summary). That's not particularly practical.
The nearby nuclear power plant here has three reactors, each of which can generate over 1100MW (one reactor is currently off-line but is on schedule to be on-line next month, now capable of up to 1280MW). Even closer to my house is the dam that can generate over 140MW.
Frequency of microwave oven as stamped on the back of the unit: 2.5GHz Wavelength of said frequency in air: About 1 foot... Are you a deluded brainless cretin as well, or simply unable to master grade-school math?
You should try some of that grade-school math (hint: units matter). The wavelength of a 2.5GHz signal is about 12cm, not 12".
Also, by your reasoning, a microwave oven could never cook (or even affect) anything smaller than 12cm, which is certainly not true. I suggest you read up on how microwave ovens work.
I have worked for small (larger than mom/pop garage but not regional/national) ISPs for over 11 years. I have seen T1 prices drop significantly in that time, but they are still a good bit more than DSL. The biggest reason is the level of service delivered. With DSL, you get "best effort" bandwidth; if the link goes down, you talk to front-line support and (mostly due to the telephone company, but again it is a cost/staffing thing) it can often take days to repair a problem. With a T1, you get your guaranteed bandwidth; if the link goes down, you talk to the network staff, and the telephone company typically must make repairs in a few hours or less (or face penalties).
Also, the hardware costs for T1 are higher. We can support something like 8000 DSL subscribers on a $25K BRAS, while a 4 port channelized DS3 card (supporting 112 T1s) runs around $45K (and that's just the interface card; the router costs another $30K+).
It is still trying to predict future results based on past performance. No matter what you predict, last year's Chipper Jones will never again face last year's Roger Clemens. Even if Clemens un-retires (again), he is not the same person, and neither is Chipper Jones. You also can't predict injuries, trades, managers' decisions, umpires' calls, weather, etc., all of which have an impact on the outcome of an individual game.
Wow, you have been to every Home Depot and Lowe's and know what they carry? I checked a local Lowe's and two local Home Depots last week and none had dimmable bulbs. I also found only one package that listed the actual lumens output instead of some made-up "like a 60W bulb". I bought incandescents instead.
It doesn't have to end up in the process memory in the clear; it could be kept in a register. Those register contents will end up on a kernel stack somewhere when you hit a process switch though. That could be limited by getting everything set up to decrypt, yeilding control back to the kernel (not sure exactly how) so that when you next run, you have a good chance at running a certain amount before being interrupted by another process. If you can decrypt in that time, the key still doesn't end up in RAM.
It is hard (and maybe impossible in this case) to do, but it could happen.
I recently downloaded a DVD image of SuSE and mailed to a friend stationed in Iraq. Another friend is only able to get dialup at his home (he could probably get satellite, but they have download caps).
High speed Internet is NOT widespread enough to require everything be done over the network. Even when it is, it is often more convenient to have media in-hand; I have more bandwidth at work (OC3+DS3s) than at home (DSL), so it sometimes still makes sense to burn things at work (or at least download to a notebook) and carry them home.
Networks are still slow: 3Mbps DSL is about the same speed as a 2x CD-ROM drive.
IIRC, the processing power can increase if you run 64 bit apps, because the 64 bit extensions include more registers (which means fewer loads and stores to RAM). 32 vs. 64 bit doesn't affect how much hard drive space you can address, as modern OSes all address >32 bit files already (unless you are still running FAT32).
Also, for the most part, 32 bit structures do NOT take up 64 bits on a 64 bit arch. An int is still 32 bits, and I believe both AMD and Intel can handle (without extra overhead) 32 bit aligned reads of 32 bit data (so no padding is required). Pointers of course are larger, but they do carry more information (with a larger virtual address space, arranging different things in RAM is easier).
Slashdot Mirror
Most people belive that means you can either accept the BSD terms or accept the GPL terms (and from then on follow only the one chosen set of terms). Theo seems to be claiming that you somehow have to follow both sets of terms. I guess it depends on your definition of "alternatively".
If you released something under the BSD license and someone made a closed-source commercial program out of it (as allowed under the BSD license and done many times by many companies), would you be incensed? If so, why would you release something under a license that allows others to do something you don't want?
The current 3 clause BSD license allows someone to release derived works under the GPL (or under closed-source commercial license). If you don't like that, then don't use the 3 clause BSD license. Licenses have specific meaning that should be understood before they are used.
Yeah, I typed that while doing something else (working on a new Nagios config) and added an extra zero.
Still, the point stands. Residential customers are not going to pay for dedicated (non-oversold) bandwidth. The costs involved in attempting to provide dedicated bandwidth for everybody would cause the prices to go up as massive amounts of equipment is replaced.
If you don't want oversold bandwidth, pull out your checkbook. The ISP I work for would be happy to sell you a 30 megabits/sec link, but it'll probably cost you around $60000/month. If you want us to "upgrade the infrastructure" to handle selling 30 megabits/sec to thousands of customers, somebody's going to have to pay that bill as well (just off the top of my head, I'd say it'll probably at least triple the cost).
We oversell DSL, but we monitor the links to make sure we have sufficient capacity for all but the most extreme peaks (the typical average bits/customer for our DSL is only around 35-40 kilobits/sec, a small fraction of each customer's available bandwidth). This is the way pretty much all modern communication networks work; land-line and cell phone networks cannot handle everybody picking up the phone and making a call at once.
At this point they'd be too far behind; it takes years to bring a new chip design to market. The current AMD and Intel chips would beat any Alpha in price/performance. I wish I could get x86 systems with the design effort of the Alpha systems though; it wasn't all about the CPU.
Not quite the same, but just last week I was over at my parents' house. We noticed the lights flickering (and the UPSes clicking) every 30-90 seconds or so for about half an hour, so we stuck a voltmeter in the socket. It was at 125V; when the lights flickered, it dropped to 120V for 30 seconds, the lights flickered again, and it went back to 125V. It did this for an hour or two before the power finally went out for an hour. The utility company had some type of failure at the substation that didn't die right away.
Yeah, but then when you get three threads, it gets more complicated. Also, one indecisive thread, and everybody stands around and complains while the one tries to decide which slice to take.
So are /. editors are just better than most at suppressing distracting stories?
I had a 4.77MHz IBM years ago. Oh wait, you said G, not M.
I was in line at the grocery store behind an astronaut (Jan Davis), but I'm in Huntsville, AL (another non-average location).
Well, the same can be said about a solar power plant, but that wasn't really my point. To replace the nuclear plant with a similar solar plant, you would have to have a solar plant that covered 125 square miles (based on the size and power listed in the summary). That's not particularly practical.
The nearby nuclear power plant here has three reactors, each of which can generate over 1100MW (one reactor is currently off-line but is on schedule to be on-line next month, now capable of up to 1280MW). Even closer to my house is the dam that can generate over 140MW.
I have worked for small (larger than mom/pop garage but not regional/national) ISPs for over 11 years. I have seen T1 prices drop significantly in that time, but they are still a good bit more than DSL. The biggest reason is the level of service delivered. With DSL, you get "best effort" bandwidth; if the link goes down, you talk to front-line support and (mostly due to the telephone company, but again it is a cost/staffing thing) it can often take days to repair a problem. With a T1, you get your guaranteed bandwidth; if the link goes down, you talk to the network staff, and the telephone company typically must make repairs in a few hours or less (or face penalties).
Also, the hardware costs for T1 are higher. We can support something like 8000 DSL subscribers on a $25K BRAS, while a 4 port channelized DS3 card (supporting 112 T1s) runs around $45K (and that's just the interface card; the router costs another $30K+).
Cleaned it out:
"Nothing for you to see here. Please move along."
It is still trying to predict future results based on past performance. No matter what you predict, last year's Chipper Jones will never again face last year's Roger Clemens. Even if Clemens un-retires (again), he is not the same person, and neither is Chipper Jones. You also can't predict injuries, trades, managers' decisions, umpires' calls, weather, etc., all of which have an impact on the outcome of an individual game.
Wow, you have been to every Home Depot and Lowe's and know what they carry? I checked a local Lowe's and two local Home Depots last week and none had dimmable bulbs. I also found only one package that listed the actual lumens output instead of some made-up "like a 60W bulb". I bought incandescents instead.
It doesn't have to end up in the process memory in the clear; it could be kept in a register. Those register contents will end up on a kernel stack somewhere when you hit a process switch though. That could be limited by getting everything set up to decrypt, yeilding control back to the kernel (not sure exactly how) so that when you next run, you have a good chance at running a certain amount before being interrupted by another process. If you can decrypt in that time, the key still doesn't end up in RAM.
It is hard (and maybe impossible in this case) to do, but it could happen.
I recently downloaded a DVD image of SuSE and mailed to a friend stationed in Iraq. Another friend is only able to get dialup at his home (he could probably get satellite, but they have download caps).
High speed Internet is NOT widespread enough to require everything be done over the network. Even when it is, it is often more convenient to have media in-hand; I have more bandwidth at work (OC3+DS3s) than at home (DSL), so it sometimes still makes sense to burn things at work (or at least download to a notebook) and carry them home.
Networks are still slow: 3Mbps DSL is about the same speed as a 2x CD-ROM drive.
Hah, the TRULY rich build a building with sufficient mass such that the Earth revolves around the building!
IIRC, the processing power can increase if you run 64 bit apps, because the 64 bit extensions include more registers (which means fewer loads and stores to RAM). 32 vs. 64 bit doesn't affect how much hard drive space you can address, as modern OSes all address >32 bit files already (unless you are still running FAT32).
Also, for the most part, 32 bit structures do NOT take up 64 bits on a 64 bit arch. An int is still 32 bits, and I believe both AMD and Intel can handle (without extra overhead) 32 bit aligned reads of 32 bit data (so no padding is required). Pointers of course are larger, but they do carry more information (with a larger virtual address space, arranging different things in RAM is easier).