* "The Soekris motherboard costs $196" It's not a motherboard, it's a single board computer. You know that and I know that, but it makes a big difference for someone not familiar with the board.
* The Mini-ITX stuff is very new, using very new technology, and I haven't seen one yet with three eth ifaces, two 'pci' slots (one mini-pci, one 3.3v pci), and two serial ports. One of the Soekris eth ifaces is net-bootable.
* The Mini-ITX stuff is fanless, but still hotter than the Soekris. I expect the Mini-ITX stuff (esp. w/ enough gear to do what the net4501 does) draws a lot more power than the 0.06A at 120V during normal (not idle) operation. Of course, the cpu on the Mini-ITX stuff is much faster. That probably doesn't matter to the target market of the net4501, because of their applications (hardware-assisted vpn and other WAN routing/firewalling tasks).
* Besides being smaller and lighter (as you mention), the net4501 runs cooler than the C3 and does not need a heat sink.
* I wonder if the Soekris has been vibration tested and such. Since the Mini-ITX stuff requires assembly (cpu, heat sink, network cards, etc), it probably couldn't withstand much vibration without some ruggedizing.
* You cannot find equivalent functionality to the net4501 for $50 anywhere (see my next paragraph =-). The net4501 is a bargain for the intended audience. And pc104 stuff is generally more expensive, because of the slightly more compact form factor and fancy i/o stuff.
I've spent a *lot* of time looking for SBCs like the net4501 but less expensive. There aren't any (yet). That's why I've got a net4501 handling many of my gateway and network tasks. For another $100 or so dollars, you can get a nice 3-port 300MHz NatSemi Geode board from Acrosser (AR-B1550, fanless and very compact) that seems adequate for file serving, mail serving, and light web serving. As with the Soekris, I haven't found better prices on compact fanless SBCs anywhere.
FWIW, the folks at Soekris and Acrosser are very pleasant to work with, and the user communities are nice.
It is only barely compliant, and only with one part of POSIX. It's not nearly enough to be genuinely useful. IIRC, Microsoft had to acheive this minimal compliance to be able to sell into certain markets (maybe the US military).
A patch has been submitted to lkml (not by me), round about 2.4.18. I sent a follow-up email to that patch today because I wanted it to get some attention. I had to do the same thing for an i810 today.
I think you are overstating the "derivative works" protection. Clean-room implementations are important for Compaq when reverse-engineering the IBM BIOS, but that seems to be a very special situation. In that case, there's only one IBM BIOS and no general body of knowledge. Furthermore, IBM has legions of lawyers and both parties have strong commercial interests.
However, if I read the GNU regex code, study the relevant automata literature, and then write my own regex parser, I do not believe my work would be considered derivative so long as I don't effectively cut-n-paste the GNU code. In this case I'm writing code from scratch using other sources as reference.
I would very much like to know which copyright cases provided precedence for this issue. Otherwise, I think the definition of "derivative" is probably up to the Librarian of Congress or similar.
Perhaps your complaint is about my use of the word "rewrite". By "rewrite" I mean write again, not fiddle with variable names and hope nobody notices.
I (personally) believe strongly in (limited-term) copyright. I don't want to discuss time limits here, but suffice it to say that I think of the GPL similarly to any other copyright license. The author asks for payment; just not in money. And the author holds the copyright and can do what she wants with that.
I like the GPL for "infrastructure". For instance, I think some of our telephone should be publically maintained; that is, Covad wouldn't have to go through Verizon to get access to the central office when installing ADSL equipment. There is a lot to think about here, and I'd want people to think about this very carefully.
I was offended by the text of the letter, let alone the purpose. I wrote to two congressmen about it because the letter was glaringly, factually wrong but was signed by leaders in our country.
A small comment on your 99% to 1% example: if a programmer is unwilling to rewrite 1% of their code in order to achieve ownership, one must assume that 1% was *extremely* valuable. If the programmer wishes to use some other license and can't replace that 1% him or herself, then he or she should start shopping for replacement code and negotiate licenses and prices with those who *can* replace that code.
Your example doesn't seem particularly strong to me. If a programmer includes *any* GPL'd code, then "paying" the "price" of respecting the GPL is "fair". Let the programmer find a cheaper/better source if they can't afford the GPL.
I agree that the National Park analogy wasn't so good, either. Maybe something like a perpetual land grant would have been better, but the underlying problem is the nature of licensing, code, and ideas. As has been requoted so often,
"If I have an apple, and give you the apple, I
have no apples. If I have an idea, and give
it to you, we both have the idea."
I don't remember the source of this quote, nor do I know if the source is truly known.
-Paul Komarek
Re:DOS didn't have automount.
on
Linux 3.0
·
· Score: 2
I've been told it's that the cdrom drivers are typically 16 bit, and that 16 bit processes use the old type of scheduling (non-preemptible).
-Paul Komarek
Re:DOS didn't have automount.
on
Linux 3.0
·
· Score: 2
My first action after installing Red Hat, every time:
1) At least twice, Torvalds had made a statement about running the best software for your job no matter what, and he made that statement in a generalized way. I want to know what his generalized idea of "best" is -- I expect it doesn't take ideals into account, and that is why he disagrees with RMS.
2) I didn't say you don't have strong ideals. I did *not* say "people not trying to replace BK have weak ideals" (that's the converse, sort of, and doesn't follow logically). I just said that the folks who would bother to replace BK with a Free alternative must have strong ideals.
I apologize for any offence from misreading my post, but there is only so much I can do about that.
RMS doesn't give a damn about convenience, especially for his own sake. He cares about ideals, and challenges everyone else to do the same. Most of his life has been spent trying to make living with his ideals more convenient. Most of us losers spend our lives rationlizing about our ideals until our ideals are convenient. I hope this explains where RMS is coming from and why he says the stuff he does.
Linus Torvalds is more like the majority of people whose ideals meld (by design or otherwise, I don't want to guess) with convenience. This is probably part of why he is fairly popular and seen as a regular guy (unlike RMS). For instance, Torvalds feels the "best" tool for the job should be used regardless. Unfortunately, Torvalds has never taken the time to describe for us exactly what "best" means. I'm sure he doesn't mean anything so sinister as "if unpaid child labor makes the tool cheaper, then it's better!", or so naive as "I'll pay anything for the best, screw value/dollar!".
RMS doesn't give a damn about Larry McVoy's company succeeding with propietary software. I believe (I don't want to put words in RMS' mouth) this is because RMS feel propietary software is worthless in the long run, and hence a waste of society's time, energy, and money. I believe he could win this argument, should he choose to make it. Torvald's feelings on the subject are useless until he defines what he means by "best".
My second personal interjection for this post is about competition versus scratching itches. It's not clear to me that the "competition" from BitKeeper is what will spur the creation of a Free package with similar features -- Torvalds doesn't seem motiviated by competition with Microsoft (he's said such several times), so it's not obvious to me that software authors will be motivated to compete with BitKeeper for the sake of competition alone.
The people in the Free and Open Source communities most affected by the lack of Free BitKeeper alternatives are the Linux (i.e. kernel) developers. Most others seem to be happy enough with existing tools, with just enough disgruntelment that subversion is finally emerging. Thus BitKeeper is providing non-Free itch-relief for the only people in the communities who are have this particular itch. Rik and Linus are dreaming if they think I care about their itches more than I care about mine. Essentially, the only people who are likely to produce BitKeeper alternatives are those kernel developers who refuse to use BitKeeper because of their ideals (if they don't use BitKeeper because they don't like source control, or because they're lazy, then they're unlikely to write a replacement).
That is, only people with strong ideals about Free software are likely to write a replacment for BitKeeper. That means people more like RMS and less like Linus.
Hear hear! I submitted a bug to Red Hat about this. They closed it after commenting that it would be better handled upstream. While that might be true, I'm still pissed. We've had virtual desktops for , what, at least 10 years? Maybe I'll go back to fvwm2 to get them back.
"The whole reason we still use pads of paper and pens is because tablet PCs aren't economically viable as an alternative -- yet."
There are other reasons, too. Weight and bulk, for instance. Also notable is the lack of foldable tablet PCs that you can wad up, put in your pocket, and maybe even still use after it goes through the laundry. Finally, I haven't seen a tablet PC that can be made into an airplane, crane, or ripped in half to share with someone lacking one.
I think the thing tablet PCs will replace is clipboards.
"But that does not provide a huge advantage for today's processors."
With the increasing disparity between cpu speed and memory latency and bandwidth, things which affect cache behavior can produce large performance gains or penalties. Dereferencing pointers at compile time is one such thing.
If the cpu-memory mismatch continues to grow, I expect that optimizers for Fortran will be able to make better use of cache than optimizers for (say) C or Java. Unless, of course, all these extensions to Fortan screw up the advantages.;-)
I don't understand your distinction between computer languages and math, especially the part about "Without an implementation". I've never seen anything suggesting that an implementation is necessary for something to be considered a "language".
You're right, and thank-you for the correction. I think I would have better stated my opinion using "engineering/applied-research" instead of "engineering". The basis for that opinion comes from complaints by researchers from JPL who don't have as much time for "research" as they would like. For me, however, JPL would be a great place to work if it weren't for the smog (thankfully the smog is decreasing).
Heh, my view of JPL is that they're a long way from pure. They seem to be as much or more an engineering-on-steroids shop than a research lab. And I think that's just what they need to be. As such, they're probably among the closest organizations in NASA to the commercial sector. JPL is in a good position to provide nearly-usable, nearly-mature technology to the relatively risk-adverse business sector.
It's not like they're selling JPL (which in fact is run by Caltech, and is not administered by the government). They're selling technology and services. That's fine with me.
Bonneville Power distributes energy from Army Core of Engineers dams to private power companies, and part of their mission as a federal organization is to keep the price as low as possible. As a result, the northwest has the cheapest electricity in nation. Furthermore, Bonneville doesn't get any tax money; their special federal privilege is to take loans from the US Treasury, which they have to pay back.
I'm sure there's plenty of other federal, state, and local governmental organizations providing services to people for a fee. For JPL to do this is an intelligent use of resources.
When I want to buy classical music (which is most of my collection), I like to go to shops that have employees who know about classical music. That eliminates every major retail outlet that doesn't specialize in classical music. And even among music specialty shops, I find that only when the owner works in the shop that I meet anyone who really knows their collection. You just can't beat a shop where the owner has listened to *everything* he or she sells.
Clean-room reverse engineering not necessary, just expedient for businesses cloning IBM's PC BIOS and such. Saying that 'unless you "clean room", your work is legally derivative' is wrong. It simply means you have to be more careful. Again, my DWT implementation is *not* derivative. Even a cursory glance at my code shows how different it is from the NRC code (and the other half dozen books I was reading at that time). A close look shows that my work is quite distinct.
Anyone *truly trying* to set a standard that *deserves to be reimplemented* is unlikely to be super picky about the subtler points of copyright law, or else they'd be defeating their purpose.
Finally, you could always use the GPL'd example in a "clean room" setting, without the reverse engineering step:
1) some engineers read the GPL'd code
2) those engineers describe the finer points of the standard (those not obvious from the high-level description) to implementation engineers.
3) check the input/output relations of the new implementation against the old implementation.
I'm not claiming that using GPL'd code in such a scenario is necessarily easier for everyone, just that it's possible if the original author desires to do so. Things would be easier if the original author forfeited copyright altogether, avoiding both the GPL *and* BSD licenses.
Do you write significant amounts of code? I'm curious, because to my eyes you (the one-or-more anonymous cowards in this thread) are oversimplifying the situation considerably. Maybe I'm being trolled; hmm.
It's a 1.2 terabyte array, and generally I end up running the fsck through a 9600 baud serial console. I have to choose between ignoring the prompts and output, or waiting a *long* time for the block counts to display. I have patience, but some things are a waste of my time.
Slashdot Mirror
Just to be picky:
* "The Soekris motherboard costs $196" It's not a motherboard, it's a single board computer. You know that and I know that, but it makes a big difference for someone not familiar with the board.
* The Mini-ITX stuff is very new, using very new technology, and I haven't seen one yet with three eth ifaces, two 'pci' slots (one mini-pci, one 3.3v pci), and two serial ports. One of the Soekris eth ifaces is net-bootable.
* The Mini-ITX stuff is fanless, but still hotter than the Soekris. I expect the Mini-ITX stuff (esp. w/ enough gear to do what the net4501 does) draws a lot more power than the 0.06A at 120V during normal (not idle) operation. Of course, the cpu on the Mini-ITX stuff is much faster. That probably doesn't matter to the target market of the net4501, because of their applications (hardware-assisted vpn and other WAN routing/firewalling tasks).
* Besides being smaller and lighter (as you mention), the net4501 runs cooler than the C3 and does not need a heat sink.
* I wonder if the Soekris has been vibration tested and such. Since the Mini-ITX stuff requires assembly (cpu, heat sink, network cards, etc), it probably couldn't withstand much vibration without some ruggedizing.
* You cannot find equivalent functionality to the net4501 for $50 anywhere (see my next paragraph =-). The net4501 is a bargain for the intended audience. And pc104 stuff is generally more expensive, because of the slightly more compact form factor and fancy i/o stuff.
I've spent a *lot* of time looking for SBCs like the net4501 but less expensive. There aren't any (yet). That's why I've got a net4501 handling many of my gateway and network tasks. For another $100 or so dollars, you can get a nice 3-port 300MHz NatSemi Geode board from Acrosser (AR-B1550, fanless and very compact) that seems adequate for file serving, mail serving, and light web serving. As with the Soekris, I haven't found better prices on compact fanless SBCs anywhere.
FWIW, the folks at Soekris and Acrosser are very pleasant to work with, and the user communities are nice.
-Paul Komarek
I believe that CF is particularly slow on writes, but not so bad on reads.
-Paul Komarek
It is only barely compliant, and only with one part of POSIX. It's not nearly enough to be genuinely useful. IIRC, Microsoft had to acheive this minimal compliance to be able to sell into certain markets (maybe the US military).
-Paul Komarek
A patch has been submitted to lkml (not by me), round about 2.4.18. I sent a follow-up email to that patch today because I wanted it to get some attention. I had to do the same thing for an i810 today.
-Paul Komarek
I think you are overstating the "derivative works" protection. Clean-room implementations are important for Compaq when reverse-engineering the IBM BIOS, but that seems to be a very special situation. In that case, there's only one IBM BIOS and no general body of knowledge. Furthermore, IBM has legions of lawyers and both parties have strong commercial interests.
However, if I read the GNU regex code, study the relevant automata literature, and then write my own regex parser, I do not believe my work would be considered derivative so long as I don't effectively cut-n-paste the GNU code. In this case I'm writing code from scratch using other sources as reference.
I would very much like to know which copyright cases provided precedence for this issue. Otherwise, I think the definition of "derivative" is probably up to the Librarian of Congress or similar.
Perhaps your complaint is about my use of the word "rewrite". By "rewrite" I mean write again, not fiddle with variable names and hope nobody notices.
-Paul Komarek
I (personally) believe strongly in (limited-term) copyright. I don't want to discuss time limits here, but suffice it to say that I think of the GPL similarly to any other copyright license. The author asks for payment; just not in money. And the author holds the copyright and can do what she wants with that.
I like the GPL for "infrastructure". For instance, I think some of our telephone should be publically maintained; that is, Covad wouldn't have to go through Verizon to get access to the central office when installing ADSL equipment. There is a lot to think about here, and I'd want people to think about this very carefully.
For "technology", I would prefer public domain.
-Paul Komarek
I was offended by the text of the letter, let alone the purpose. I wrote to two congressmen about it because the letter was glaringly, factually wrong but was signed by leaders in our country.
-Paul Komarek
A small comment on your 99% to 1% example: if a programmer is unwilling to rewrite 1% of their code in order to achieve ownership, one must assume that 1% was *extremely* valuable. If the programmer wishes to use some other license and can't replace that 1% him or herself, then he or she should start shopping for replacement code and negotiate licenses and prices with those who *can* replace that code.
Your example doesn't seem particularly strong to me. If a programmer includes *any* GPL'd code, then "paying" the "price" of respecting the GPL is "fair". Let the programmer find a cheaper/better source if they can't afford the GPL.
I agree that the National Park analogy wasn't so good, either. Maybe something like a perpetual land grant would have been better, but the underlying problem is the nature of licensing, code, and ideas. As has been requoted so often,
"If I have an apple, and give you the apple, I
have no apples. If I have an idea, and give
it to you, we both have the idea."
I don't remember the source of this quote, nor do I know if the source is truly known.
-Paul Komarek
I've been told it's that the cdrom drivers are typically 16 bit, and that 16 bit processes use the old type of scheduling (non-preemptible).
-Paul Komarek
My first action after installing Red Hat, every time:
rpm -e magicdev
-Paul Komarek
Suddenly I'm very happy that Python prints stack traces in reverse, so that the non-bullshit part doesn't scroll of the top of the screen.
-Paul Komarek
1) At least twice, Torvalds had made a statement about running the best software for your job no matter what, and he made that statement in a generalized way. I want to know what his generalized idea of "best" is -- I expect it doesn't take ideals into account, and that is why he disagrees with RMS.
2) I didn't say you don't have strong ideals. I did *not* say "people not trying to replace BK have weak ideals" (that's the converse, sort of, and doesn't follow logically). I just said that the folks who would bother to replace BK with a Free alternative must have strong ideals.
I apologize for any offence from misreading my post, but there is only so much I can do about that.
-Paul Komarek
RMS doesn't give a damn about convenience, especially for his own sake. He cares about ideals, and challenges everyone else to do the same. Most of his life has been spent trying to make living with his ideals more convenient. Most of us losers spend our lives rationlizing about our ideals until our ideals are convenient. I hope this explains where RMS is coming from and why he says the stuff he does.
Linus Torvalds is more like the majority of people whose ideals meld (by design or otherwise, I don't want to guess) with convenience. This is probably part of why he is fairly popular and seen as a regular guy (unlike RMS). For instance, Torvalds feels the "best" tool for the job should be used regardless. Unfortunately, Torvalds has never taken the time to describe for us exactly what "best" means. I'm sure he doesn't mean anything so sinister as "if unpaid child labor makes the tool cheaper, then it's better!", or so naive as "I'll pay anything for the best, screw value/dollar!".
RMS doesn't give a damn about Larry McVoy's company succeeding with propietary software. I believe (I don't want to put words in RMS' mouth) this is because RMS feel propietary software is worthless in the long run, and hence a waste of society's time, energy, and money. I believe he could win this argument, should he choose to make it. Torvald's feelings on the subject are useless until he defines what he means by "best".
My second personal interjection for this post is about competition versus scratching itches. It's not clear to me that the "competition" from BitKeeper is what will spur the creation of a Free package with similar features -- Torvalds doesn't seem motiviated by competition with Microsoft (he's said such several times), so it's not obvious to me that software authors will be motivated to compete with BitKeeper for the sake of competition alone.
The people in the Free and Open Source communities most affected by the lack of Free BitKeeper alternatives are the Linux (i.e. kernel) developers. Most others seem to be happy enough with existing tools, with just enough disgruntelment that subversion is finally emerging. Thus BitKeeper is providing non-Free itch-relief for the only people in the communities who are have this particular itch. Rik and Linus are dreaming if they think I care about their itches more than I care about mine. Essentially, the only people who are likely to produce BitKeeper alternatives are those kernel developers who refuse to use BitKeeper because of their ideals (if they don't use BitKeeper because they don't like source control, or because they're lazy, then they're unlikely to write a replacement).
That is, only people with strong ideals about Free software are likely to write a replacment for BitKeeper. That means people more like RMS and less like Linus.
-Paul Komarek
Hear hear! I submitted a bug to Red Hat about this. They closed it after commenting that it would be better handled upstream. While that might be true, I'm still pissed. We've had virtual desktops for , what, at least 10 years? Maybe I'll go back to fvwm2 to get them back.
-Paul Komarek
Don't worry, I'm sure President Bozo has already added suicidal cells to the axis of evil.
-Paul Komarek
"The whole reason we still use pads of paper and pens is because tablet PCs aren't economically viable as an alternative -- yet."
There are other reasons, too. Weight and bulk, for instance. Also notable is the lack of foldable tablet PCs that you can wad up, put in your pocket, and maybe even still use after it goes through the laundry. Finally, I haven't seen a tablet PC that can be made into an airplane, crane, or ripped in half to share with someone lacking one.
I think the thing tablet PCs will replace is clipboards.
-Paul Komarek
I'm not so sure about your last claim:
;-)
"But that does not provide a huge advantage for today's processors."
With the increasing disparity between cpu speed and memory latency and bandwidth, things which affect cache behavior can produce large performance gains or penalties. Dereferencing pointers at compile time is one such thing.
If the cpu-memory mismatch continues to grow, I expect that optimizers for Fortran will be able to make better use of cache than optimizers for (say) C or Java. Unless, of course, all these extensions to Fortan screw up the advantages.
-Paul Komarek
Parts 1 and 2 go together. Fortran prevents certain "pointer nightmares" allowed in C which reduce performance.
-Paul Komarek
I don't understand your distinction between computer languages and math, especially the part about "Without an implementation". I've never seen anything suggesting that an implementation is necessary for something to be considered a "language".
-Paul Komarek
You're right, and thank-you for the correction. I think I would have better stated my opinion using "engineering/applied-research" instead of "engineering". The basis for that opinion comes from complaints by researchers from JPL who don't have as much time for "research" as they would like. For me, however, JPL would be a great place to work if it weren't for the smog (thankfully the smog is decreasing).
-Paul Komarek
Heh, my view of JPL is that they're a long way from pure. They seem to be as much or more an engineering-on-steroids shop than a research lab. And I think that's just what they need to be. As such, they're probably among the closest organizations in NASA to the commercial sector. JPL is in a good position to provide nearly-usable, nearly-mature technology to the relatively risk-adverse business sector.
It's not like they're selling JPL (which in fact is run by Caltech, and is not administered by the government). They're selling technology and services. That's fine with me.
Bonneville Power distributes energy from Army Core of Engineers dams to private power companies, and part of their mission as a federal organization is to keep the price as low as possible. As a result, the northwest has the cheapest electricity in nation. Furthermore, Bonneville doesn't get any tax money; their special federal privilege is to take loans from the US Treasury, which they have to pay back.
I'm sure there's plenty of other federal, state, and local governmental organizations providing services to people for a fee. For JPL to do this is an intelligent use of resources.
-Paul Komarek
When I want to buy classical music (which is most of my collection), I like to go to shops that have employees who know about classical music. That eliminates every major retail outlet that doesn't specialize in classical music. And even among music specialty shops, I find that only when the owner works in the shop that I meet anyone who really knows their collection. You just can't beat a shop where the owner has listened to *everything* he or she sells.
-Paul Komarek
Clean-room reverse engineering not necessary, just expedient for businesses cloning IBM's PC BIOS and such. Saying that 'unless you "clean room", your work is legally derivative' is wrong. It simply means you have to be more careful. Again, my DWT implementation is *not* derivative. Even a cursory glance at my code shows how different it is from the NRC code (and the other half dozen books I was reading at that time). A close look shows that my work is quite distinct.
Anyone *truly trying* to set a standard that *deserves to be reimplemented* is unlikely to be super picky about the subtler points of copyright law, or else they'd be defeating their purpose.
Finally, you could always use the GPL'd example in a "clean room" setting, without the reverse engineering step:
1) some engineers read the GPL'd code
2) those engineers describe the finer points of the standard (those not obvious from the high-level description) to implementation engineers.
3) check the input/output relations of the new implementation against the old implementation.
I'm not claiming that using GPL'd code in such a scenario is necessarily easier for everyone, just that it's possible if the original author desires to do so. Things would be easier if the original author forfeited copyright altogether, avoiding both the GPL *and* BSD licenses.
Do you write significant amounts of code? I'm curious, because to my eyes you (the one-or-more anonymous cowards in this thread) are oversimplifying the situation considerably. Maybe I'm being trolled; hmm.
-Paul Komarek
I had to use HPUX in front of a client (research sponser) on a locked-down PA-RISC machine to fix bugs on-site. That was scary enough for me. ;-)
-Paul Komarek
It's a 1.2 terabyte array, and generally I end up running the fsck through a 9600 baud serial console. I have to choose between ignoring the prompts and output, or waiting a *long* time for the block counts to display. I have patience, but some things are a waste of my time.
-Paul Komarek