On the issue of liability, how would you make that work? If I create a program and you choose to use it, you want me to be responsible for fixing any problems with the program for as long as you use it? Or can I declare it "end of life"? If so, what prevents me from putting it out there, freely distributable, and then declaring it obsolete/"end of life" right away? Or if I cannot decide when to stop supporting it, what happens when my program develops a problem (or rather, the problem is discovered) years after I released it? I might have moved on, died, forgotten how it worked, or some other reason might make it impossible for me to fix it. How about my expenses? What if I'm employed and cannot fix the problem in the time I have (or perhaps I'm not even allowed, for contractual reasons).
If you want some kind of liability, that's fine by me, but you will have to pay me an amount that reasonably covers my expenses with regard to this liability. So if you get a program from me for free, or even for a small fee, don't expect me to fix problems that require much time or cost money to fix (I might do so anyway, but that's a different matter).
And if you need me to be liable for problems, why are you using software that does not come with a warranty?
Sure, but the issue of classpath lacking behind the official java class library seems to be a non-issue, as the official class library is available anyway.
True. But for starters it is not free (in the GNU sense). I also think that Sun would be very much against (an likely within their rights to be) anyone using their classes seperately, at least in a distribution. And then there is the issue of classes that need to be written specifically for the VM (interface to classloader and such), or tied closely to native code (like AWT/Swing, it needs to talk to the native GUI).
As for documentation, people might have differing preferences, but to me the documentation is actually the best thing about java. I've never needed to buy a single Java book. No other language I have seen so far has matched the documentation that is available for Java.
Last time I checked, C# documentation was available in either Word.doc-format or in windows help format - YUK!
No argument there, as a developer, the documentation for java is mostly nice (I can't recall them, but I think there are/has been places where the documentation was lacking, but it might have been my understanding of the big picture). But I'm just not sure all of the corner cases are specified, and that makes it harder to build a work-alike.
If you look in the Blackdown FAQ (link to first section), you will see that Blackdown is a port of java to linux, not an opensource version. They are quite clear on being a port of Suns JDK, and while I would have loved to try it out, I have never come across a source release of Blackdowns java.
You are however right that it is entirely possible to have an opensource implementation of java, like GNU classpath and gcc's gcj (class library and compiler/vm respectively). But just like WINE always lacks behind Windows, classpath is likely to lack behind the official java class library, especially if documentation is lacking (not something I can judge, but developing in java I do seem to recall occations where the API documentation from Sun was not too good, and I was only trying to use it - not trying to recreate it).
If you jump through Suns hoops, you can indeed download the source for the JDK. It's not easy to build it, but it can be done (I've had it running some versions ago). It is just not under an open license, so you can't really do much with it, except build it and run it.
That is the default behaviour of Windows. It is however possible to seperate the mice by inserting a filter in the driver-stack. Unfortunately this requires support from applications that wish to use it. CPNMouse is a library that implements a basic api for such support.
One could implement it in such a way that the validity of the keys overlap. Then one could read the key, perhaps even keeping it "locked" on the device for the time it takes to enter it, and then be sure that it is still valid (unless you take a very long time entering it). There would need to ba an overlap in any case, since the network transmission could cause the key to be too old by the time it got to the server.
But if it is implemented in some soft-/hardware combination, then those timing issues are much smaller. But then why not just use a big-ass public key-pair.
No, because your provider may have added features to his version and licensed it to you under a proprietary license. Those features will not be in the code you can download, or the code may have disappeared from the net. Either way you cannot alter the software while retaining the features your provider added.
If the original was GPL, then any features your provider wanted to add but not GPL would have to be put into a seperate program (or seperated from the GPL'ed code in some other way, so as to not be derivative), which could be used by your improved version of the GPL'ed code.
So in either case you get two things: The original program and the extra features. But only in the GPL case can you be assured that you can alter the program without loosing the extra features (which were presumably good, since you wanted them enough to agree to the proprietary license).
I'm not usually one to point out other peoples mistakes, but this is just ridiculous - The movie is called 'Wedlock', and it is right there at the top of the IMDb page.
Pretty much where I was coming from. I have also messed around, hacking around broken stuff, with both Imake and autoconf/automak based stuff, and Imake just looks worse (maybe it would be different if I had some experience with either, but just fixing a broken link parameter and that sort of thing seems much simpler with auto*). And if it gets the options of X into a configure script instead the host.def file, then I wont complain (force of habit I guess).
If I read it correctly, they are replacing the Imake buildsystem with GNU autoconf/automake - a good thing I think. Other than that I don't know what the differences are (going to be), but they will probably try to integrate some of the other parts of the freedesktop platform.
I have no idea about his situation, but I know that some houses here (in Denmark) are protected (as historic buildings), and as such the things you can do to them are limited. In that case it might make perfect sence to use wireless, if you want to avoid having cables lying around. Or maybe he just wanted to have the connection hidden and found that having proper hidden cabling done (in his home) would be more expensive (and/or otherwise undesirable) than doing a wireless link.
How do you come to that conclusion? HTML defines how the site may describe their content. If they answer your request with relatively valid HTML (or some other content-type that your browser understands) they've done exactly what you asked them to. If your chosen browser then request additional content to show what the HTML described, and they comply, how are they at fault? The ads are annoying and all that, but you asked for them (or at least caused your browser to do so) so go use another browser if you are unsatisfied.
I also tried to view their demo (IE with googlebar), but all that happened was that googlebar blocked one popup, and their questioneer loaded without insident. If that's the best they've got, let them have their fun.
Or do anyone have a link to a real page on one of the mentioned sites, where one might see their technology in action?
If you use this approach you may want to run a small webserver on your machine (needs only to be available to 127.0.0.1), and have it return something usefull, like a 1x1 pixel transparent image as 404 reply. That way most pages should look the same (if they size their images, and they should really do that), and the browser gets a reply right away instead of having to figure out the fact that your machine is not answering on port 80 (might not take that long, but some browsers might spend a little time on it).
You also don't know how many bytes the flash intro, the menu bottons, and the background is going to consume. Are you deciding what content I'm allowed to put on my site? I'm not very interested in having ads shoved in my face, but when you follow a link on the web you accept that your browser will get the html file, and if you use a reasonably standard browser it will also get the graphical content referenced by that page. Sorry, that's the rules of the game, and if you don't like it, you have to use something other than a standard browser (perhaps a popup-blocking browser+a proxy, a non-javascript browser, or maybe a text only browser). But whatever I decide to put on my page is my content, and whether your browser chooses to get it alongside the page is your problem.
Another way these nonsense spams work is, in my experience, by having two different MIME parts, a plaintext part of random words, and an html part with the actual spam content. Since I don't use html mail, it works rather poorly on me, but I did once take a look at the html part, and it was formated text, not random nonsense like in the plaintext part.
One thing I think you fail to see, is that on the list of places to be colonized Mars is high on the list (since there are not many places on earth that are yet uninhabited (the bottom of the ocean maybe, but that is also not without risk)). One required part of a colonization effort is the actual colonists, which in this case would be the astrounauts. They would then explore, build, and learn how to live on Mars (or die trying), and as the project progress they might be joined by additional voluntiers hitching a ride on the biannual supply rocket.
While I'm not personally enough of an explorer type to want to go on those conditions, I think that the concept is fine. And given the things people sacrifice their lives for, I think it would be a fine cause. And I think that the decision about which odds the astrounauts are willing to risk should be left to them, let society discuss whether the mission gives benefits proportional to the resources spent (count the ACTUAL cost of the astrounauts lives if you will, but unless you want to stop any progress of the human race, you need to accept that every step forward caries a cost lives (sometimes a potential risk, other times people WILL die to make a better world for all of us)). Plus - you are forgetting that if the astrounauts survive long enough they might get a ride home on the spaceships that are developed using the knowledges gained from getting the astrounauts out there in the first place. Or maybe they will call home and ask for some chemical processing equipment to be put into the next resupply rocket, and start the first gas-station outside the earths gravitywell. Then it is not a true one-way trip, it is just a dangourous voyage with a high risk of not coming home, something that explorers and soldiers have dealt with since the dawn of time.
In short: Let the people who are putting their lives on the line decide what odds they will take, and what potential gain they think is worth their lives.
Isn't that the idea of discovery: You go to court with what little you have, and then you have the judge compel the other side to produce records and such, just like IBM is doing to SCO (OK, SCO are dragging their feet).
That can be done with a single connection in the chip package (which can be altered after testing), I seem to recall seeing guides to how one could use a pencil to change the identity of processors (perhaps some Pentium II or III's?), allowing higher speed or whatever. And besides, just because the original stencil is the same, you are not sure to end up with the same chip in the end, unless the chip has been tested to the standard desired.
So while an XP might just need a few wires crossed to identify as an MP and act like one, AMD has not found it to live up to the standard of an MP (SMP does require more than single processor, the two processors have to agree on timings and such), and thus they refuse to sell it as such.
If they also allow some MP grade processors to go into the XP bin for marketing reasons, then that's AMDs business, you are free not to buy there products. And the XP/MP case is a matter of getting more than you pay for, not less (an MP is an MP, but a cheaper XP may be an MP in disguise).
Since the license for gtk+ is LGPL there should be no problems linking a closed source app against gtk+ and QT (under whatever license Trolltech wants to sell you), provided that you can license the integration library under compatible terms (LGPL or license from author). If you can't, you can go and do your own integration library, I don't think anybody will stop you.
DAldredge: I'm not assuming one way or the other about your personal views, but these kinds of questions often carries a kind of unspoken assumsion that licensing your library under the GPL takes away the rights of others. My view is that it does not, it just does not grant as many privileges as for instance the LGPL, but whether to grant them must be left up to the author of the library.
Seems strange if AOL's own nameserver doesn't know about it. Note that the second time around I'm asking dns-01.ns.aol.com, which is mentioned in the SOA entry of aol.com. And just to add, I've also tried the other AOL nameservers. Where is this fantastic SPF entry?
You just boot from a floppy or CD with at least modular support for the controller, compile the new kernel and install. As 2.6.x becomes the default for for distros, the drivers will either be compiled in or put in initrd on the installer kernels, same as all other unusual storage device drivers. How is this different from SCSI controlers or IDE-RAID contollers?
If your favorite distro have 2.6.x but doesn't have an ISO or floppy out with the SATA driver available, ask them. Is that too hard?
Or do you want the kernel, the vanilla kernel.org kernel, to force me to compile every SATA driver (and to be fair: every NIC, SCSI, IDE, etc. driver), just to insure that it will be there for you to install with? I'm fairly sure (but no kernel handy, so I can't check) that you could compile a kernel without drivers for standard IDE controllers, there just ain't many good reasons to do so on x86.
Slashdot Mirror
On the issue of liability, how would you make that work? If I create a program and you choose to use it, you want me to be responsible for fixing any problems with the program for as long as you use it? Or can I declare it "end of life"? If so, what prevents me from putting it out there, freely distributable, and then declaring it obsolete/"end of life" right away? Or if I cannot decide when to stop supporting it, what happens when my program develops a problem (or rather, the problem is discovered) years after I released it? I might have moved on, died, forgotten how it worked, or some other reason might make it impossible for me to fix it. How about my expenses? What if I'm employed and cannot fix the problem in the time I have (or perhaps I'm not even allowed, for contractual reasons).
If you want some kind of liability, that's fine by me, but you will have to pay me an amount that reasonably covers my expenses with regard to this liability. So if you get a program from me for free, or even for a small fee, don't expect me to fix problems that require much time or cost money to fix (I might do so anyway, but that's a different matter).
And if you need me to be liable for problems, why are you using software that does not come with a warranty?
True. But for starters it is not free (in the GNU sense). I also think that Sun would be very much against (an likely within their rights to be) anyone using their classes seperately, at least in a distribution. And then there is the issue of classes that need to be written specifically for the VM (interface to classloader and such), or tied closely to native code (like AWT/Swing, it needs to talk to the native GUI).
No argument there, as a developer, the documentation for java is mostly nice (I can't recall them, but I think there are/has been places where the documentation was lacking, but it might have been my understanding of the big picture). But I'm just not sure all of the corner cases are specified, and that makes it harder to build a work-alike.
If you look in the Blackdown FAQ (link to first section), you will see that Blackdown is a port of java to linux, not an opensource version. They are quite clear on being a port of Suns JDK, and while I would have loved to try it out, I have never come across a source release of Blackdowns java.
You are however right that it is entirely possible to have an opensource implementation of java, like GNU classpath and gcc's gcj (class library and compiler/vm respectively). But just like WINE always lacks behind Windows, classpath is likely to lack behind the official java class library, especially if documentation is lacking (not something I can judge, but developing in java I do seem to recall occations where the API documentation from Sun was not too good, and I was only trying to use it - not trying to recreate it).
If you jump through Suns hoops, you can indeed download the source for the JDK. It's not easy to build it, but it can be done (I've had it running some versions ago). It is just not under an open license, so you can't really do much with it, except build it and run it.
That is the default behaviour of Windows. It is however possible to seperate the mice by inserting a filter in the driver-stack. Unfortunately this requires support from applications that wish to use it. CPNMouse is a library that implements a basic api for such support.
One could implement it in such a way that the validity of the keys overlap. Then one could read the key, perhaps even keeping it "locked" on the device for the time it takes to enter it, and then be sure that it is still valid (unless you take a very long time entering it). There would need to ba an overlap in any case, since the network transmission could cause the key to be too old by the time it got to the server.
But if it is implemented in some soft-/hardware combination, then those timing issues are much smaller. But then why not just use a big-ass public key-pair.
No, because your provider may have added features to his version and licensed it to you under a proprietary license. Those features will not be in the code you can download, or the code may have disappeared from the net. Either way you cannot alter the software while retaining the features your provider added.
If the original was GPL, then any features your provider wanted to add but not GPL would have to be put into a seperate program (or seperated from the GPL'ed code in some other way, so as to not be derivative), which could be used by your improved version of the GPL'ed code.
So in either case you get two things: The original program and the extra features. But only in the GPL case can you be assured that you can alter the program without loosing the extra features (which were presumably good, since you wanted them enough to agree to the proprietary license).
Sorry about that, totally forgot to check for aliases, since I _knew_ that I was right - stupid thing to do when you are wrong.
I'm not usually one to point out other peoples mistakes, but this is just ridiculous - The movie is called 'Wedlock', and it is right there at the top of the IMDb page.
Pretty much where I was coming from. I have also messed around, hacking around broken stuff, with both Imake and autoconf/automak based stuff, and Imake just looks worse (maybe it would be different if I had some experience with either, but just fixing a broken link parameter and that sort of thing seems much simpler with auto*). And if it gets the options of X into a configure script instead the host.def file, then I wont complain (force of habit I guess).
Another one is X-Win32 from StarNet. Works pretty well, and runs root-less (I think it can also run 'in a window').
If I read it correctly, they are replacing the Imake buildsystem with GNU autoconf/automake - a good thing I think. Other than that I don't know what the differences are (going to be), but they will probably try to integrate some of the other parts of the freedesktop platform.
I have no idea about his situation, but I know that some houses here (in Denmark) are protected (as historic buildings), and as such the things you can do to them are limited. In that case it might make perfect sence to use wireless, if you want to avoid having cables lying around. Or maybe he just wanted to have the connection hidden and found that having proper hidden cabling done (in his home) would be more expensive (and/or otherwise undesirable) than doing a wireless link.
How do you come to that conclusion? HTML defines how the site may describe their content. If they answer your request with relatively valid HTML (or some other content-type that your browser understands) they've done exactly what you asked them to. If your chosen browser then request additional content to show what the HTML described, and they comply, how are they at fault? The ads are annoying and all that, but you asked for them (or at least caused your browser to do so) so go use another browser if you are unsatisfied.
I also tried to view their demo (IE with googlebar), but all that happened was that googlebar blocked one popup, and their questioneer loaded without insident. If that's the best they've got, let them have their fun.
Or do anyone have a link to a real page on one of the mentioned sites, where one might see their technology in action?
If you use this approach you may want to run a small webserver on your machine (needs only to be available to 127.0.0.1), and have it return something usefull, like a 1x1 pixel transparent image as 404 reply. That way most pages should look the same (if they size their images, and they should really do that), and the browser gets a reply right away instead of having to figure out the fact that your machine is not answering on port 80 (might not take that long, but some browsers might spend a little time on it).
You also don't know how many bytes the flash intro, the menu bottons, and the background is going to consume. Are you deciding what content I'm allowed to put on my site? I'm not very interested in having ads shoved in my face, but when you follow a link on the web you accept that your browser will get the html file, and if you use a reasonably standard browser it will also get the graphical content referenced by that page. Sorry, that's the rules of the game, and if you don't like it, you have to use something other than a standard browser (perhaps a popup-blocking browser+a proxy, a non-javascript browser, or maybe a text only browser). But whatever I decide to put on my page is my content, and whether your browser chooses to get it alongside the page is your problem.
Another way these nonsense spams work is, in my experience, by having two different MIME parts, a plaintext part of random words, and an html part with the actual spam content. Since I don't use html mail, it works rather poorly on me, but I did once take a look at the html part, and it was formated text, not random nonsense like in the plaintext part.
One thing I think you fail to see, is that on the list of places to be colonized Mars is high on the list (since there are not many places on earth that are yet uninhabited (the bottom of the ocean maybe, but that is also not without risk)). One required part of a colonization effort is the actual colonists, which in this case would be the astrounauts. They would then explore, build, and learn how to live on Mars (or die trying), and as the project progress they might be joined by additional voluntiers hitching a ride on the biannual supply rocket.
While I'm not personally enough of an explorer type to want to go on those conditions, I think that the concept is fine. And given the things people sacrifice their lives for, I think it would be a fine cause. And I think that the decision about which odds the astrounauts are willing to risk should be left to them, let society discuss whether the mission gives benefits proportional to the resources spent (count the ACTUAL cost of the astrounauts lives if you will, but unless you want to stop any progress of the human race, you need to accept that every step forward caries a cost lives (sometimes a potential risk, other times people WILL die to make a better world for all of us)). Plus - you are forgetting that if the astrounauts survive long enough they might get a ride home on the spaceships that are developed using the knowledges gained from getting the astrounauts out there in the first place. Or maybe they will call home and ask for some chemical processing equipment to be put into the next resupply rocket, and start the first gas-station outside the earths gravitywell. Then it is not a true one-way trip, it is just a dangourous voyage with a high risk of not coming home, something that explorers and soldiers have dealt with since the dawn of time.
In short: Let the people who are putting their lives on the line decide what odds they will take, and what potential gain they think is worth their lives.
Isn't that the idea of discovery: You go to court with what little you have, and then you have the judge compel the other side to produce records and such, just like IBM is doing to SCO (OK, SCO are dragging their feet).
That can be done with a single connection in the chip package (which can be altered after testing), I seem to recall seeing guides to how one could use a pencil to change the identity of processors (perhaps some Pentium II or III's?), allowing higher speed or whatever. And besides, just because the original stencil is the same, you are not sure to end up with the same chip in the end, unless the chip has been tested to the standard desired.
So while an XP might just need a few wires crossed to identify as an MP and act like one, AMD has not found it to live up to the standard of an MP (SMP does require more than single processor, the two processors have to agree on timings and such), and thus they refuse to sell it as such.
If they also allow some MP grade processors to go into the XP bin for marketing reasons, then that's AMDs business, you are free not to buy there products. And the XP/MP case is a matter of getting more than you pay for, not less (an MP is an MP, but a cheaper XP may be an MP in disguise).
Since the license for gtk+ is LGPL there should be no problems linking a closed source app against gtk+ and QT (under whatever license Trolltech wants to sell you), provided that you can license the integration library under compatible terms (LGPL or license from author). If you can't, you can go and do your own integration library, I don't think anybody will stop you.
DAldredge: I'm not assuming one way or the other about your personal views, but these kinds of questions often carries a kind of unspoken assumsion that licensing your library under the GPL takes away the rights of others. My view is that it does not, it just does not grant as many privileges as for instance the LGPL, but whether to grant them must be left up to the author of the library.
Seems strange if AOL's own nameserver doesn't know about it. Note that the second time around I'm asking dns-01.ns.aol.com, which is mentioned in the SOA entry of aol.com. And just to add, I've also tried the other AOL nameservers. Where is this fantastic SPF entry?
Anybody know why I just get this:
[dossen@horse09:~]$ dig aol.com txt
; <<>> DiG 9.2.1 <<>> aol.com txt
;; global options: printcmd
;; Got answer:
;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 39946
;; flags: qr rd ra; QUERY: 1, ANSWER: 0, AUTHORITY: 1, ADDITIONAL: 0
;; QUESTION SECTION:
;aol.com. IN TXT
;; AUTHORITY SECTION:
aol.com. 20 IN SOA dns-01.ns.aol.com. hostmaster.aol.net. 2004010902 1800 300 604800 600
;; Query time: 3 msec
;; SERVER: 130.225.16.40#53(130.225.16.40)
;; WHEN: Fri Jan 9 16:39:20 2004
;; MSG SIZE rcvd: 89
And if I ask AOL themselves:
[dossen@horse09:~]$ dig aol.com txt @dns-01.ns.aol.com
; <<>> DiG 9.2.1 <<>> aol.com txt @dns-01.ns.aol.com
;; global options: printcmd
;; Got answer:
;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 1932
;; flags: qr aa rd; QUERY: 1, ANSWER: 0, AUTHORITY: 1, ADDITIONAL: 0
;; QUESTION SECTION:
;aol.com. IN TXT
;; AUTHORITY SECTION:
aol.com. 600 IN SOA dns-01.ns.aol.com. hostmaster.aol.net. 2004010902 1800 300 604800 600
;; Query time: 119 msec
;; SERVER: 152.163.159.232#53(dns-01.ns.aol.com)
;; WHEN: Fri Jan 9 16:40:45 2004
;; MSG SIZE rcvd: 89
You just boot from a floppy or CD with at least modular support for the controller, compile the new kernel and install. As 2.6.x becomes the default for for distros, the drivers will either be compiled in or put in initrd on the installer kernels, same as all other unusual storage device drivers. How is this different from SCSI controlers or IDE-RAID contollers?
If your favorite distro have 2.6.x but doesn't have an ISO or floppy out with the SATA driver available, ask them. Is that too hard?
Or do you want the kernel, the vanilla kernel.org kernel, to force me to compile every SATA driver (and to be fair: every NIC, SCSI, IDE, etc. driver), just to insure that it will be there for you to install with? I'm fairly sure (but no kernel handy, so I can't check) that you could compile a kernel without drivers for standard IDE controllers, there just ain't many good reasons to do so on x86.