"Whereas a competent MCSE or IT director will have properly secured a corporation's machines against remote exploits (a properly designed network, even if none of the machines had been patched, should've been able to stay free of worms like Blaster and Welchia, for example)"
Indeed. Most workplaces will strip search their employees for floppies, CD's, USB disks, and impound any laptops employees attempt to bring to work. VPN or dialup connections to the corporate LAN are absolutely forbidden. No employees are to recieve any mail with any form of attachments in your average corporation.
Or not.
Face it, firewalls and various other protective schemes in the network at best slow down the attack. It's not even close to a sure way to avoid getting hit. Any competent IT director understands that perimeter protection is mainly useful for preventing intrusions.
"Windows NT-based operating systems listen on so many ports, and are designed so wide open, because they are meant to sit inside a secured corporate network."
No. Windows NT-based operating systems listen on so many ports and are designed so wide open because MS as a company is completely incompetent when it comes to security. There are no 'secured corporate networks', because that would mean unhooking from the internet and forbidding any way to connect to, or transfer data into the network.
"If I had used the courts to enforce the GPL years ago, Microsoft's whispering would now be falling on deaf ears. Just this month I have been working on a couple of moderately sticky situations. ``Look,'' I say, ``at how many people all over the world are pressuring me to enforce the GPL in court, just to prove I can. I really need to make an example of someone. Would you like to volunteer?''" - Eben Moglen, Professor of Law and Legal history, General Counsel of the Free Software Foundation
Sounds like he's got his volunteer to be made an example of.
Before complaining about the previous commenters reading of the article, did you read more than the blurb at the top of the page?
'While viruses containing mouse IL-4 should not be lethal to humans, recombinant viruses can have unexpected effects, he says. "You'd hope the combination remains mouse-specific."'
and
'Why his group's engineered viruses are not contagious is a mystery, he says.'
It sounds like the scientists are not quite agreed on wether or not it may or may not be either contagious or dangerous to humans.
"So, if you have to transmit information to two other computers for every disk write, you're effectively limitting yourself to a maximum of about 5 megabytes/second disk transfer."
That's just an implementation issue. I'd suggest using multicast.
I rather agree that there are better ways to solve the issues around distributed data storage than to write all servers at once tho. Performance isnt the biggest issue, integrity and syncronization would be worse.
"Unless national security laws prohibit distribution of sources to places where binary distributions are permitted, there is no conflict with the GPL."
There still is no conflict with the GPL; the resolution to any such conflict is built into it.
The GPL clearly states that if you are unable to fulfil the terms of the license for any other legal reason, then you may not perform any distribution at all.
That means if security laws prohibit distribution of sources somewhere then the GPL prohibits distribution of binaries there as well. No conflict occurs.
"This lawsuit is - if (as looks likely?) it goes to trial - going to be make-or-break time for the GPL."
The GPL is constructed to defeat any attempt to challange it. It's not a make-or-break time, it's a make-or-make time.
If the GPL is 'defeated', then SCO's license to distribute the code is automatically revoked and they can only use it as far as copyright allows them. They have no other license. It's GPL or copyright, period. The GPL doesnt go beyond copyright in any way, which means that getting it declared void merely gives you less rights.
"There's no way, no how that they could write a volume manager or filesystem product that's even in the same league with VxFS and VxVM."
There are already several replacements for VxFS which, while useful, is not very unique. LVM also works as a replacement for VxVM, and there are other alternatives worked on. The alternatives to Veritas also have the huge advantage that they are free and integrated in Linux, which makes a huge difference in what level of PITA it becomes when dealing with it.
I'll gladly pay $2500 for a RedHat AS support contract, but I'm not, _ever_, going back to 'please enter the license key code for your filesystem'.
Um, if you have discovered that Veritas is not very well designed why do you want it more integrated?
Veritas is like that. Either you get to live with it or you take a hard long look at the more free replacements.
You can live without Veritas today. I most definitely dont want it included in RedHat. The alternatives like LVM are far more worthwhile to pursue (And more in line with RedHat's tendency to prefer freely distributable software in the distribution. Which is one of the main reasons that RedHat has a far bigger marketshare than SuSE does).
You're definitely not alone. I want a phone I can reliably use for making phonecalls. The way things seem to be going phonecalls are starting to be a race between wether the battery will run out before the PhoneOS crashes.
Soon they'll be able to run the cellphone switches on Windows 95, because frankly nobody would notice if the whole switch goes dead, people will just sit around powering their phones off and on again.
Oh, well, it's not like Goldstar is a quality brand anyway.
"GNU libc leads all libc implementations by a large margin in bloat and waste of memory. One day it got so painful that I wrote my own libc. With this, a static binary of 'Hello world' took only 300 bytes..."
So, did his Hello World support multibyte character sets, or, in fact, any sort of internationalization?
Sure libc is hugely bloated. But most programs link to it dynamically so it will be loaded once. Imagine the bloat if libc was ultraslim and each and every application implemented the bloat on their own. Bloat that would not be shared between programs. Not to mention the amount of work repeated.
"But the Free Software Foundation doesn't want royalties -- it wants you to burn down your house, or at the very least share it with cloners."
That line is rather funny. He seems to have missed the point that, in fact, it's NOT Linksys's house. Linksys only built the outhouse, and the FSF wants Linksys to share the outhouse in exchange for Linksys being allowed to use the mansion. Not to mention that Linksys remains entirely free to take their outhouse and remove it from the FSF property, as long as they quit using the mansion and build their own mansion.
One link about the Lindows NDA would be: http://old.lwn.net/2002/0418/. The companies in question were Lindows and Corel. You may be able to find more on google; unfortunatly it seems it's much easier to find references to the problems than references to the peaceful resolutions.
As these cases concern distributions, as software collections, it's not entirely an exact match for the case either; as collections carry a separate copyright on the collection in itself, it's possible to slap an NDA on the collection itself. The GPL terms would be fulfilled as long as any GPL components in the collection were not covered by the NDA and the source to those components were released. As it's hard to find good references on the resolution of the cases I cant tell if that's how it was dealt with.
Anyways, I think you're right that the most damaging aspect of such issues would be the negative publicity anyway. I dont think it's possible to sidestep the GPL by hiding organizational borders under an NDA, but the decision on wether distribution has or has not taken place would be up to a court. There are even murkier areas there tho; what if you sign a consultancy contract with your prospective customers? Or what about cross-ownership between companies? In such cases it might come down to intent.
Personally I doubt IBM would go down the proprietary Unix road again. They've learned the lesson that "proprietary and different" in general means "worse" in the Unix world; the resulting lack of ports and supported software means you lose more buisness than you gain through any proprietary nice features. Better for them to gain customers through competent consulting and good hardware. The only Unix company that didnt quite get that part is probably Sun, but they've gotten away with it by being the most common platform.
Well, first of all, if your company rewrites half of Apache you can do whatever you want with the code except call it Apache, if I remember the Apache license correctly. The Apache web server is not under the GPL.
Second, an NDA does not a legal organizational entity make. You can try signing an NDA with a 'development partner' and give them Windows under your site license and see what MS thinks about wether you're the same organization or not.
Patent law has nothing to do with it. The companies may be partners however much they wish to; they're still not the same organization, not when contracting Windows, not when distributing GPL software, nor when sued for copyright violations.
"We are developing software. We think we can get patents. We cannot discuss the features that might be patented with anybody with whom we do not have an NDA without losing our ability to patent those features."
The restrictions on patents have nothing to do with distribution between organizations. The restrictions are that there must not be any previously published prior art (altho with the current patent offices around the world you could probably publish it in 200 meter burning letters across the sky and they wouldnt notice anyway). You are, of course, entirely able to limit this publishing by enforcing an NDA, which works fine with patents, as patent restrictions arent concerned with limited distribution, only publication, and as a consequence of that, with public distribution. Copyrights on the other hand are concerned with nothing but distribution, in any way or form. The legal aspects are very different.
As an aside, it actually has happened (twice, I believe). The companies in question engaging in the NDA covered distribution changed the NDA's and published source after chats with FSF's legal counsel, even if they were just temporarily engaging in the practice for understandable reasons.
"If you give or sell it to another company, you must make the source available to that company, but you could contract that they not distribute the software or the source to anybody else. An NDA (Non-Disclosure Agreement) allows you to share technology without allowing that technology to be distributed to others. A well-written NDA could be used to sell GPL'd code without providing source code."
No, limiting the recievers rights is not acceptable under the GPL. Section 6: "You may not impose any further restrictions on the recipients' exercise of the rights granted herein.". This applies to any kind of limitations, ranging from patents to NDA's.
The GPL is pretty well thought through. There are no loopholes like that. The only grey areas are in the areas of dynamic runtime loading of code to specified ABI's and the application service provider hole.
"SGI has remedied the *MINOR* breaches it previously had, and is free. Its own analysis proved that."
Well, according to the usual definition of copyright they would be. However, SCO isnt using the same definition of derivative works as the rest of us. Remember, they're claiming that everything like XFS or JFS or RCU that has been developed to be able to run on something based on SCO's intellectual property is by extension the property of SCO too.
So if you rewrite the law and live in an imaginary world then SCO is correct and IBM and SGI arent in the clear.
But you know, all those valid uses could be done on a Commodore 64 or similar ancient cheap computer. In fact, a specifically designed computer for the purpose could probably be mass produced for less than $10 these days...
"Isn't Michigan having a budget crunch like every other state?"
Exactly. So the Michigan "school superintendents and educational information technology leaders" cant pay for their own "low-key gourmet dinner at the very upscale Tribute restaurant in Farmington Hills". Enter Dell. The 6th graders dont need any laptops, but the superintendents sure cant go without those gourmet dinners.
All of these 'lets spend the tax payers money on junk that will become obsolete before it's useful and half of the students will probably break' ideas are a complete waste of money. I didnt hear anything about getting a free typewriter or calculator or encyclopedia when I was in 6th grade. Things that might have been as, if not more, useful.
If they think kids in 6th grade need some form of computer education it would be far more productive to give them a Commodore 64 and teach them Basic or something. Cheaper, easier to understand for a beginner if 'understanding computers' is the goal rather than 'using computers', and far less easy to break and abuse for games and pr0n surfing.
I have only one piece of advice for David Strom. He should read about the product before commenting.
It's painfully obvious that the Java Desktop System is just Linux with Gnome, and has very little to do with Java. Which sortof makes Mr Stroms article more or less completely beside the point and makes it clear that he doesnt know what he's talking about.
It's an understandable error to make, as apparently Suns Product Naming Division had some difficulty too. Still, one would think a journalist would read more than the name of the product before writing a column.
Well, I'm looking forward to his next rant... about why Microsoft should stick to writing software instead of branching into glass construction work with that 'Windows' product.
"Publishing a list of words and phrases that are considered unacceptable to a specific belief, opinion, or legal system very closely resembles censorship."
Quote from the FSF site: "There are a number of words and phrases which we recommend avoiding, or avoiding in certain contexts and usages. The reason is either that they are ambiguous, or that they imply an opinion that we hope you may not entirely agree with."
In my opinion, that doesnt sound like censorship at all. It's merely a recommendations on words to avoid because the words are not suitable and confuse the issues.
Using a terminology that minimises the risks of confusion is always a good idea if you're interested in avoiding misunderstandings.
Many of the terms the FSF recommends avoiding are confusing in the extreme. IP is one of the absolutely worst of them, as it both contains inherent propaganda that attempts to draw paralells between physical property rights and limited rights to ideas and it's close to worthless as a grouping as the laws it attempts to include are far too diverse to be considered as a whole.
Actually, I dont think it's hard to pair program. The difficult part of pair programming is finding the right combination of pairs. Stick the wrong people together and you get a pair that is less efficient combined than either would be on their own. Stick the right pair together and you get more done than both of them working separately.
As long as the DNS specification is extended in a normal fashion through the IETF that would be perfectly acceptable.
As it is, DNS already has different query types for MX, NS and A records. Extending with more query types would not be technically impossible. Name caching would, of course, have to be done separately on a record type basis as an http request might not generate the same response as a ssh request (for example, for a http related request it might make sense to return multiple ip adresses to try for loadbalance and redundancy reasons, which might not make sense for ssh).
Improving the wheel is not the same as reinventing the wheel. Reinventing the wheel means someone has already invented the wheel, but still you try to reinvent the same wheel because it's not _your_ wheel/you've misunderstood why the wheel is round/etc.
The X wheel is already round. Going through the phases of making triangular wheels, square wheels and octagonal wheels until arriving at the round wheel, again, seems redundant.
I have yet to see a let's-replace-X proposal that has succeeded in making a convincing case it would end up in any way better. This one doesnt either.
"It would be interesting if instead of typing in a mispelled name......it redirected you to a google search. I wonder how it would be received if they did that."
I'm perfectly happy with them redirecting to their own site if that's what they want. If, and _only_ if, they make the service optional and ensure it doesnt interfere with DNS lookups pertaining to any other protocol than http.
Most objections arent about what they do with their changes, the objections are that they change anything at all when there are many technical assumptions about a specific behaviour. Changing this breaks a lot of things.
This functionality belongs in the browser or in an extension to DNS. Not as a sudden destructive change in the.com and.net domains.
"VHS macrovision is popular precisely because it's undetectable in how it alters visual quality. You'll hear lots of complaints by people who are unable to copy videos correctly, but you'll never hear a complaint by anyone about how macrovision has degraded their signal -- it hasn't."
Of course you wont hear anyone complaining about how macrovision has degraded their signal - most people dont know about macrovision and would never use the expression 'degraded their signal' when talking about why their picture quality is shit.
I had to junk an older but perfectly working TV due to macrovision screwing picture quality on it, and I've had to buy a clunky switch to manually switch between VCR and DVD because of fscking macrovision prevents me from routing the DVD through the VCR. Eventually I'll get a real signal amplifier to fix it.
Once you explain to people what macrovision is and what it does you'll get all sorts of 'oooh. aaah. So that's why my store-bought Disney tapes look like shit'. They usually have been thinking their video/tv/whatever was broken or that the tape was of bad quality.
Macrovision isnt even close to undetectable. The only reason it doesnt result in mass returns is that people think something's wrong with them or their TV/VCR.
A WX type record would probably make more sense to implement slightly differently. Note that it wouldnt become an implementation issue for the browsers until they'd actually decide to resolve the new type of records instead of old type records, and it should be possible to resolve both types during a migration phase (or permanently).
Replying with a list of adresses would actually make sense for web browsers, and it would be IMO a more elegant solution to some forms of load balancing and redundancy than we often see today. It would only reply with multiple records when there are multiple servers serving the exact same site, of course, but preference would make sense for this type of records, just as with mail to allow both fallback and loadbalancing.
When you'd request domain.com you'd get the main site for the domain. If you request foo.domain.com, you'd get foo.domain.com, unless foo.domain.com does not exist in which case you'd either get a no-such-host or, optionally depending on how domain.com owners configure it, you'd get a wildcard pointing to either domain.com or a sitemap for domain.com, for example.
So, basically, it should serve the same purpose as A records, but extended for loadbalancing and redundancy purposes, and to allow wildcarding for http requests without affecting other protocols.
Slashdot Mirror
"Whereas a competent MCSE or IT director will have properly secured a corporation's machines against remote exploits (a properly designed network, even if none of the machines had been patched, should've been able to stay free of worms like Blaster and Welchia, for example)"
Indeed. Most workplaces will strip search their employees for floppies, CD's, USB disks, and impound any laptops employees attempt to bring to work. VPN or dialup connections to the corporate LAN are absolutely forbidden. No employees are to recieve any mail with any form of attachments in your average corporation.
Or not.
Face it, firewalls and various other protective schemes in the network at best slow down the attack. It's not even close to a sure way to avoid getting hit. Any competent IT director understands that perimeter protection is mainly useful for preventing intrusions.
"Windows NT-based operating systems listen on so many ports, and are designed so wide open, because they are meant to sit inside a secured corporate network."
No. Windows NT-based operating systems listen on so many ports and are designed so wide open because MS as a company is completely incompetent when it comes to security. There are no 'secured corporate networks', because that would mean unhooking from the internet and forbidding any way to connect to, or transfer data into the network.
Well, it's gonna make Eben Moglen happy at least;
"If I had used the courts to enforce the GPL years ago, Microsoft's whispering would now be falling on deaf ears. Just this month I have been working on a couple of moderately sticky situations. ``Look,'' I say, ``at how many people all over the world are pressuring me to enforce the GPL in court, just to prove I can. I really need to make an example of someone. Would you like to volunteer?''" - Eben Moglen, Professor of Law and Legal history, General Counsel of the Free Software Foundation
Sounds like he's got his volunteer to be made an example of.
Before complaining about the previous commenters reading of the article, did you read more than the blurb at the top of the page?
'While viruses containing mouse IL-4 should not be lethal to humans, recombinant viruses can have unexpected effects, he says. "You'd hope the combination remains mouse-specific."'
and
'Why his group's engineered viruses are not contagious is a mystery, he says.'
It sounds like the scientists are not quite agreed on wether or not it may or may not be either contagious or dangerous to humans.
"So, if you have to transmit information to two other computers for every disk write, you're effectively limitting yourself to a maximum of about 5 megabytes/second disk transfer."
That's just an implementation issue. I'd suggest using multicast.
I rather agree that there are better ways to solve the issues around distributed data storage than to write all servers at once tho. Performance isnt the biggest issue, integrity and syncronization would be worse.
"Unless national security laws prohibit distribution of sources to places where binary distributions are permitted, there is no conflict with the GPL."
There still is no conflict with the GPL; the resolution to any such conflict is built into it.
The GPL clearly states that if you are unable to fulfil the terms of the license for any other legal reason, then you may not perform any distribution at all.
That means if security laws prohibit distribution of sources somewhere then the GPL prohibits distribution of binaries there as well. No conflict occurs.
"This lawsuit is - if (as looks likely?) it goes to trial - going to be make-or-break time for the GPL."
The GPL is constructed to defeat any attempt to challange it. It's not a make-or-break time, it's a make-or-make time.
If the GPL is 'defeated', then SCO's license to distribute the code is automatically revoked and they can only use it as far as copyright allows them. They have no other license. It's GPL or copyright, period. The GPL doesnt go beyond copyright in any way, which means that getting it declared void merely gives you less rights.
"There's no way, no how that they could write a volume manager or filesystem product that's even in the same league with VxFS and VxVM."
There are already several replacements for VxFS which, while useful, is not very unique. LVM also works as a replacement for VxVM, and there are other alternatives worked on. The alternatives to Veritas also have the huge advantage that they are free and integrated in Linux, which makes a huge difference in what level of PITA it becomes when dealing with it.
I'll gladly pay $2500 for a RedHat AS support contract, but I'm not, _ever_, going back to 'please enter the license key code for your filesystem'.
Um, if you have discovered that Veritas is not very well designed why do you want it more integrated?
Veritas is like that. Either you get to live with it or you take a hard long look at the more free replacements.
You can live without Veritas today. I most definitely dont want it included in RedHat. The alternatives like LVM are far more worthwhile to pursue (And more in line with RedHat's tendency to prefer freely distributable software in the distribution. Which is one of the main reasons that RedHat has a far bigger marketshare than SuSE does).
You're definitely not alone. I want a phone I can reliably use for making phonecalls. The way things seem to be going phonecalls are starting to be a race between wether the battery will run out before the PhoneOS crashes.
Soon they'll be able to run the cellphone switches on Windows 95, because frankly nobody would notice if the whole switch goes dead, people will just sit around powering their phones off and on again.
Oh, well, it's not like Goldstar is a quality brand anyway.
"GNU libc leads all libc implementations by a large margin in bloat and waste of memory. One day it got so painful that I wrote my own libc. With this, a static binary of 'Hello world' took only 300 bytes..."
So, did his Hello World support multibyte character sets, or, in fact, any sort of internationalization?
Sure libc is hugely bloated. But most programs link to it dynamically so it will be loaded once. Imagine the bloat if libc was ultraslim and each and every application implemented the bloat on their own. Bloat that would not be shared between programs. Not to mention the amount of work repeated.
I dont think that would really be an improvement.
"But the Free Software Foundation doesn't want royalties -- it wants you to burn down your house, or at the very least share it with cloners."
That line is rather funny. He seems to have missed the point that, in fact, it's NOT Linksys's house. Linksys only built the outhouse, and the FSF wants Linksys to share the outhouse in exchange for Linksys being allowed to use the mansion. Not to mention that Linksys remains entirely free to take their outhouse and remove it from the FSF property, as long as they quit using the mansion and build their own mansion.
One link about the Lindows NDA would be: http://old.lwn.net/2002/0418/. The companies in question were Lindows and Corel. You may be able to find more on google; unfortunatly it seems it's much easier to find references to the problems than references to the peaceful resolutions.
As these cases concern distributions, as software collections, it's not entirely an exact match for the case either; as collections carry a separate copyright on the collection in itself, it's possible to slap an NDA on the collection itself. The GPL terms would be fulfilled as long as any GPL components in the collection were not covered by the NDA and the source to those components were released. As it's hard to find good references on the resolution of the cases I cant tell if that's how it was dealt with.
Anyways, I think you're right that the most damaging aspect of such issues would be the negative publicity anyway. I dont think it's possible to sidestep the GPL by hiding organizational borders under an NDA, but the decision on wether distribution has or has not taken place would be up to a court. There are even murkier areas there tho; what if you sign a consultancy contract with your prospective customers? Or what about cross-ownership between companies? In such cases it might come down to intent.
Personally I doubt IBM would go down the proprietary Unix road again. They've learned the lesson that "proprietary and different" in general means "worse" in the Unix world; the resulting lack of ports and supported software means you lose more buisness than you gain through any proprietary nice features. Better for them to gain customers through competent consulting and good hardware. The only Unix company that didnt quite get that part is probably Sun, but they've gotten away with it by being the most common platform.
"So my company rewrites half of Apache."
Well, first of all, if your company rewrites half of Apache you can do whatever you want with the code except call it Apache, if I remember the Apache license correctly. The Apache web server is not under the GPL.
Second, an NDA does not a legal organizational entity make. You can try signing an NDA with a 'development partner' and give them Windows under your site license and see what MS thinks about wether you're the same organization or not.
Patent law has nothing to do with it. The companies may be partners however much they wish to; they're still not the same organization, not when contracting Windows, not when distributing GPL software, nor when sued for copyright violations.
"We are developing software. We think we can get patents. We cannot discuss the features that might be patented with anybody with whom we do not have an NDA without losing our ability to patent those features."
The restrictions on patents have nothing to do with distribution between organizations. The restrictions are that there must not be any previously published prior art (altho with the current patent offices around the world you could probably publish it in 200 meter burning letters across the sky and they wouldnt notice anyway). You are, of course, entirely able to limit this publishing by enforcing an NDA, which works fine with patents, as patent restrictions arent concerned with limited distribution, only publication, and as a consequence of that, with public distribution. Copyrights on the other hand are concerned with nothing but distribution, in any way or form. The legal aspects are very different.
As an aside, it actually has happened (twice, I believe). The companies in question engaging in the NDA covered distribution changed the NDA's and published source after chats with FSF's legal counsel, even if they were just temporarily engaging in the practice for understandable reasons.
"If you give or sell it to another company, you must make the source available to that company, but you could contract that they not distribute the software or the source to anybody else. An NDA (Non-Disclosure Agreement) allows you to share technology without allowing that technology to be distributed to others. A well-written NDA could be used to sell GPL'd code without providing source code."
No, limiting the recievers rights is not acceptable under the GPL. Section 6: "You may not impose any further restrictions on the recipients' exercise of the rights granted herein.". This applies to any kind of limitations, ranging from patents to NDA's.
The GPL is pretty well thought through. There are no loopholes like that. The only grey areas are in the areas of dynamic runtime loading of code to specified ABI's and the application service provider hole.
"SGI has remedied the *MINOR* breaches it previously had, and is free. Its own analysis proved that."
Well, according to the usual definition of copyright they would be. However, SCO isnt using the same definition of derivative works as the rest of us. Remember, they're claiming that everything like XFS or JFS or RCU that has been developed to be able to run on something based on SCO's intellectual property is by extension the property of SCO too.
So if you rewrite the law and live in an imaginary world then SCO is correct and IBM and SGI arent in the clear.
But you know, all those valid uses could be done on a Commodore 64 or similar ancient cheap computer. In fact, a specifically designed computer for the purpose could probably be mass produced for less than $10 these days...
"Isn't Michigan having a budget crunch like every other state?"
Exactly. So the Michigan "school superintendents and educational information technology leaders" cant pay for their own "low-key gourmet dinner at the very upscale Tribute restaurant in Farmington Hills". Enter Dell. The 6th graders dont need any laptops, but the superintendents sure cant go without those gourmet dinners.
All of these 'lets spend the tax payers money on junk that will become obsolete before it's useful and half of the students will probably break' ideas are a complete waste of money. I didnt hear anything about getting a free typewriter or calculator or encyclopedia when I was in 6th grade. Things that might have been as, if not more, useful.
If they think kids in 6th grade need some form of computer education it would be far more productive to give them a Commodore 64 and teach them Basic or something. Cheaper, easier to understand for a beginner if 'understanding computers' is the goal rather than 'using computers', and far less easy to break and abuse for games and pr0n surfing.
I have only one piece of advice for David Strom. He should read about the product before commenting.
It's painfully obvious that the Java Desktop System is just Linux with Gnome, and has very little to do with Java. Which sortof makes Mr Stroms article more or less completely beside the point and makes it clear that he doesnt know what he's talking about.
It's an understandable error to make, as apparently Suns Product Naming Division had some difficulty too. Still, one would think a journalist would read more than the name of the product before writing a column.
Well, I'm looking forward to his next rant... about why Microsoft should stick to writing software instead of branching into glass construction work with that 'Windows' product.
"Publishing a list of words and phrases that are considered unacceptable to a specific belief, opinion, or legal system very closely resembles censorship."
Quote from the FSF site: "There are a number of words and phrases which we recommend avoiding, or avoiding in certain contexts and usages. The reason is either that they are ambiguous, or that they imply an opinion that we hope you may not entirely agree with."
In my opinion, that doesnt sound like censorship at all. It's merely a recommendations on words to avoid because the words are not suitable and confuse the issues.
Using a terminology that minimises the risks of confusion is always a good idea if you're interested in avoiding misunderstandings.
Many of the terms the FSF recommends avoiding are confusing in the extreme. IP is one of the absolutely worst of them, as it both contains inherent propaganda that attempts to draw paralells between physical property rights and limited rights to ideas and it's close to worthless as a grouping as the laws it attempts to include are far too diverse to be considered as a whole.
"it's hard to pair program"
Actually, I dont think it's hard to pair program. The difficult part of pair programming is finding the right combination of pairs. Stick the wrong people together and you get a pair that is less efficient combined than either would be on their own. Stick the right pair together and you get more done than both of them working separately.
As long as the DNS specification is extended in a normal fashion through the IETF that would be perfectly acceptable.
As it is, DNS already has different query types for MX, NS and A records. Extending with more query types would not be technically impossible. Name caching would, of course, have to be done separately on a record type basis as an http request might not generate the same response as a ssh request (for example, for a http related request it might make sense to return multiple ip adresses to try for loadbalance and redundancy reasons, which might not make sense for ssh).
Improving the wheel is not the same as reinventing the wheel. Reinventing the wheel means someone has already invented the wheel, but still you try to reinvent the same wheel because it's not _your_ wheel/you've misunderstood why the wheel is round/etc.
The X wheel is already round. Going through the phases of making triangular wheels, square wheels and octagonal wheels until arriving at the round wheel, again, seems redundant.
I have yet to see a let's-replace-X proposal that has succeeded in making a convincing case it would end up in any way better. This one doesnt either.
"It would be interesting if instead of typing in a mispelled name......it redirected you to a google search. I wonder how it would be received if they did that."
.com and .net domains.
I'm perfectly happy with them redirecting to their own site if that's what they want. If, and _only_ if, they make the service optional and ensure it doesnt interfere with DNS lookups pertaining to any other protocol than http.
Most objections arent about what they do with their changes, the objections are that they change anything at all when there are many technical assumptions about a specific behaviour. Changing this breaks a lot of things.
This functionality belongs in the browser or in an extension to DNS. Not as a sudden destructive change in the
"VHS macrovision is popular precisely because it's undetectable in how it alters visual quality. You'll hear lots of complaints by people who are unable to copy videos correctly, but you'll never hear a complaint by anyone about how macrovision has degraded their signal -- it hasn't."
Of course you wont hear anyone complaining about how macrovision has degraded their signal - most people dont know about macrovision and would never use the expression 'degraded their signal' when talking about why their picture quality is shit.
I had to junk an older but perfectly working TV due to macrovision screwing picture quality on it, and I've had to buy a clunky switch to manually switch between VCR and DVD because of fscking macrovision prevents me from routing the DVD through the VCR. Eventually I'll get a real signal amplifier to fix it.
Once you explain to people what macrovision is and what it does you'll get all sorts of 'oooh. aaah. So that's why my store-bought Disney tapes look like shit'. They usually have been thinking their video/tv/whatever was broken or that the tape was of bad quality.
Macrovision isnt even close to undetectable. The only reason it doesnt result in mass returns is that people think something's wrong with them or their TV/VCR.
A WX type record would probably make more sense to implement slightly differently. Note that it wouldnt become an implementation issue for the browsers until they'd actually decide to resolve the new type of records instead of old type records, and it should be possible to resolve both types during a migration phase (or permanently).
Replying with a list of adresses would actually make sense for web browsers, and it would be IMO a more elegant solution to some forms of load balancing and redundancy than we often see today. It would only reply with multiple records when there are multiple servers serving the exact same site, of course, but preference would make sense for this type of records, just as with mail to allow both fallback and loadbalancing.
When you'd request domain.com you'd get the main site for the domain. If you request foo.domain.com, you'd get foo.domain.com, unless foo.domain.com does not exist in which case you'd either get a no-such-host or, optionally depending on how domain.com owners configure it, you'd get a wildcard pointing to either domain.com or a sitemap for domain.com, for example.
So, basically, it should serve the same purpose as A records, but extended for loadbalancing and redundancy purposes, and to allow wildcarding for http requests without affecting other protocols.