It's easy - in fact, my comment gives just about all the description you need. Obviously, as with any method, you can write pages and pages about the consequences for strategic voting etc, but my reading on the topic has only pointed out one disadvantage: you have to rank everyone you approve of equally. Which is sort of a natural consequence of choosing to keep the ballot simple and not having ranked ballots.
Basically, strategic voting in an Approval system is a matter of choosing where to "draw the line" in your own preferences to decide who to vote for or not. For example, if my true preferences are Nader > Kerry > Bush, I have to decide whether I'm going to vote for Nader+Kerry or just Nader. That choice depends on how far apart the candidates are in my opinion, and how likely I think they are to win.
The ability/need to make this decision does have some arguably negative consequences (especially compared to Concordet which can reasonably claim to be entirely immune to strategy), but IMHO nothing compared to the negative consequences of all the other options, including Concordet (which has the IMHO fatal flaw that the ballot and counting processes are both too hard for the average voter).
I disagree that this would turn a state with three electoral votes into a battleground.
Consider that in order to actually get all three votes, you'd need to have a very large proportion of the popular vote (depending on how you round, somewhere between 75% and 85%). Then bear in mind that if you have *any* hope of achieving that proportion, you can probably guarantee at least one, if not two, of the electoral votes with no campaigning at all. Now you're fighting over one or two votes again.
Contrast to a state with 25 electoral votes, where even though you have to persuade more voters, you only have to shift 4%-5% of the voters statewide to get the same one vote as you do from Vermont.
I'm sure it's easier to effect a shift of opinion in 4% of a very large population than in 85% of a small population.
But the main point is that if *all* the states are battlegrounds, the candidates have to have broad appeal and not just to those in a few large states with close to 50-50 polling.
Same reason that I support Approval voting: "broad appeal" is exactly the feature a president should have!
2) "Just like IRV" means that you require the whole thirty second soundbite explaining IRV to happen first, so you exceeded your thirty seconds that way. Actually I think IRV will fail based on that criterion too - thirty seconds is longer than any quote I've seen on the news from any of the presidential candidates, or on any other topic for that matter. I can't remember seeing anything as complex as IRV *ever* explained on the news.
3) Approval voting: "Just like today except you get to vote for as many candidates as you like". That's less than a *5* second soundbite. Why go to all the trouble of explaing IRV in the first place, when (a) it sucks and (b) approval is so much simpler to explain?
Better than eliminating the electoral college would be to require each state to allocate its electors proportionately instead of winner-take-all for the state.
That would completely eliminate the concept of a "battleground state" as it exists now, and "florida" situations in the future - there would never be a situation where a small increase in real votes could net you 21 electoral votes in one shot. Any recounts would be, at most, fighting over one electoral vote at a time instead of a whole state's worth, because the margin of error is never so large that it would cover more than that proportion of the state's voters.
I think this would probably have to be federally or constitutionally mandated, because individual states that apply it to only themselves instantly *dis*advantage themselves: where they might previously have gotten lots of attention from the candidates because 20+ electoral votes were up for grabs, the candidates would now concentrate on the states that *hadn't* implemented the change.
Concordet is clearly complex and despite having spent some time trying to understand it, I still don't fully grok it.
But how the heck is Approval complex? Anyone who's ever used a computer UI can have it explained in a single sentence "The ballot's a list of checkboxes instead of radio buttons". Even without that, it's pretty easy to say "Just check off as many names as you like", or "It's just like what we have now except you can vote for more than one person".
It's actually *simpler* than the current system, because even the morons who voted Buchanan/Gore in 2000 would have had their votes counted!
On the other hand, again having researched this a little bit, I still have no idea how you would construct a paper ballot for IRV that *I* could understand easily, let alone those people. IRV seems to practically *demand* electronic voting machines - and we all know what the state of the art in those is like!
You need to require that the licensee grants to *everyone*, not just the licensor, a right to use all their patents used in the same product as the licensed patent (I'd like to say "all patents owned by the same person" but that's probably impractical) under either the terms of the "PGPL" or under completely unrestricted terms.
This is more like the actual behavior of the GPL: As soon as you combine GPL code with something else into a single product, you're required to distribute the whole thing under the GPL. Similarly, this proposed license would say that as soon as you use this patent along with some other patented technique in the same product, you have to release the other patent under the same terms.
Now, if IBM and Novell could both be persuaded to license their entire patent pools under terms like these, we might be getting somewhere.
There's one big gotcha that I can see: As long as you only require the "grant-back" for patents owned by the licensee, there's a loophole where the licensee simply sells their patents to someone else prior to doing the licensing, and then gets revenue from them. On the other hand, if you require it for all patents, owned or not, then you have the problem of code that inadvertently infringes one of the zillions of bogus patents - logically, this should be permitted, but under the license it wouldn't be.
Re:Seems extremely difficult with chemical rockets
on
After the X Prize
·
· Score: 1
Question: Isn't SS1 effectively a multi-stage vehicle, with White Knight being the first stage?
The key insight in SS1 appears to me that you can make the throwaway stages also function as manned, landable aircraft. If you made a much bigger version of White Knight, Mounted under that a much bigger SS1, and attached to *that* a normal-sized SS1, wouldn't you effectively have a three-stage fully reusable orbit-ready vehicle? Essentially you launch the third stage from the top of the current SS1 hop.
Obviously I'm fully aware that the problems are nowhere near as simple as I'm implying. Quite apart from the fact that you can't simply scale up SS1 or WK and expect them to continue working, SS1 would probably not work very well as a carrier (that's why WK looks so different than SS1) and WK could certainly not feather in the way that SS1 does to get back from the separation point so probably all three vehicles would need to be radically different from the current WK and SS1. But I don't see any problem in principle.
Firefox supports a whitelist of sites that you can xpinstall from. This was added in the Preview Release, I believe. If you look in the release notes of that version, there should be more information on the whitelist and how to change its contents. Emptying the whitelist will effectively disable installing extensions.
What's astonishing to me is that *nobody* mentions that the only reason he's so far ahead of previous records is that in all previous seasons you were booted after 5 wins. Of *course* someone doing very well is going to beat records by a lot - anyone who did this well on any previous season would have left the show three weeks ago with a paltry 150K or so.
Sure, the guy's good. But there's no way to tell how much better he is than previous 5-day winners, because those previous winners never got the chance to show what they could do.
If I were single the answer would absolutely be "Hell yeah!".
Since I'm married with a daughter and a son on the way, I'd have to say no - unless I could persuade them to come with me.
I'd love to emigrate to Mars. If the option becomes available within my lifetime, I *know* I'll try to persuade them. My wife won't go for it, but who knows how my kids will feel when they grow up... maybe they can persuade Mommy;)
I just tested a gnome-terminal that I happened to have open, and click-drag-to-select-menu-item worked fine. I don't have the latest GNOME, but that makes your statement suspect to me. Which version of GNOME were you using when you saw this behavior?
So use Konsole or gnome-terminal (okay, not gnome-terminal if you value the speed of your shell - they *really* need to apply the year-old patch to fix the scrolling speed - but I'm sure there are other alternatives as well) which has 'copy' and 'paste' in the menus...
In every program I use in Linux (specifically, Mozilla, rdesktop, and various GNOME stuff) Ctrl-C/Ctrl-V work perfectly, and so does the select/middleclick system, and neither interferes with each other.
I haven't used KDE in a long time but I understand that they introduced the same behavior with Qt3/KDE3.
Unless you're using really ancient software, pretty much everything will work in *either* mode, or you can do what I do and use a combination of both (choosing whether to bother pressing Ctrl-C to copy depending on whether you're going to need to highlight something at the destination).
I'm really curious to understand how so many people manage to still have a problem with this. Are you perhaps expecting that since "everybody knows that select copies on Linux", Ctrl-V will paste the thing that you last selected, instead of the last thing you Ctrl-C'd, and not testing it to verify this? Or just assuming that selecting something will overwrite your Ctrl-C buffer? I'd like to believe that people would actually test these things before posting Ask Slashdots about it, but you have to wonder...
Yeah, as I said in the original post, I've written code (and a naive computer player, which I wrote to mitigate the difficulty of finding other people to play against) to play it using a grid rendering like my poor-quality ascii-art rendition, in HTML tables.
I still have the code but it's a horrible perl hack with dependencies on software I'm reluctant to release out of sheer shame, and I don't have a public server to host it on.
I keep intending to rewrite it under Mono or something, but never get around to it. The AI is really simple but still quite hard to beat. I suppose now I have to try your suggestion of playing it in a 5^4 grid instead of 4^4...
Not sure if you're being tongue-in-cheek here, but if you're not and have actually played 5^4 tictactoe, what's it like to play? Is there a noticeable advantage to getting the center square?
4D tictactoe (naughts and crosses, for brits) is a very difficult game to master. I've written an implementation with a fairly naive computer player that just weights each square based on the sum of a weighting of each line that goes through it, and I still have a really hard time beating the computer at it. I guess the computer's just much better at visualizing 4 dimensions than I am;)
If you're interested in the implementation, I'm afraid it's not publically available right now, but it's not that hard to write. The main insight is that your entire 4D gui can be done in straight HTML tables (let's see if I can get/. to let me render this... I'll show a 2^4 grid instead of the normal 4^4 one to save typing...)
|_|_| |_|_| |_|_| |_|_|
|_|_| |_|_| |_|_| |_|_|
Why 4^4 rather than 3^4? It turns out that 3^n for any n>2 has an easy strategy that allows the first player to always win. Proving that is left as an exercise for the reader.
My proposed Open Source iTunes store would honor the DRM. Of course, being Open Source it's impossible to stop someone else from coming along and taking that DRM code out again, but how many average users would bother to recompile an application that came with their distribution, when for the most part it would let them do what they wanted?
Remember, for most people burning to a CD and ripping it (which Apple allows, so my proposed app would too) is much easier than downloading a patch and recompiling the app. And of the people who would go to this trouble, how many of them wouldn't just go and get Playfair off some P2P network if they were using the Apple client instead? The existence of an Open Source client that honors the DRM is no worse than the existence of Apple's client with Playfair out of the bag (which it is, and it ain't going back in).
The point is that even if the actual app we build honors the DRM, the player itself has to "break" it to get to the content to allow it to be played. So, paradoxically, we need the DRM to be broken before we can write an Open Source app that honors it.
So many people are missing one important reason why PlayFair is important.
Think "iTunes store for Linux".
Think Open Source iTunes store for Linux.
Think someone reverse-engineering the protocol for iTunes store and allowing music to be downloaded - AND PAID FOR - on a Linux computer using only Open Source software.
That would be a big deal. Apple's never going to do it. Playfair is necessary (but not sufficient) to make it happen.
Furthermore, it's probably the only way that I'd ever use iTunes store. I haven't decided for sure whether I'd consider using a binary proprietary iTunes release for Linux, but I certainly won't be using a Windows or Mac version.
The Open Source client could even enforce the DRM when used unmodified - I'd still use it, and I wouldn't hack the source to remove it. Probably some people would, but those people already have Playfair anyway. I have no need to bypass those restrictions, but I do need a Linux client if I'm going to use the thing at all.
So in a small way, Playfair would allow Apple to make a little more money (by selling to Linux users - a small but not non-existent, and growing market) without costing them a cent in development costs.
Red Hat, eh? I'm surprised you don't get "apt-get: command not found", but I guess times change...
Heh, I doubt that adding debian unstable to your sources-list would help much - dpkg is probably not included with RH9:) A quick google doesn't reveal any good kaffe RPM sources - perhaps the Kaffe developers all use different distributions.
If you're comfortable building from source, the tarball from kaffe.org shouldn't cause any problems.
I heard that Red Hat is pushing gcj fairly heavily as a solution, so I would expect that gij is either present by default on RH9 or easy to install using whatever normal means Red Hat users use for getting software.
(Btw, Kaffe is merging its class libraries with GNU Classpath and is able to run many current apps, such as Tomcat and Eclipse. The version in debian unstable is (at long last) pretty current. I don't know what the debian status of gcj/gij is. Work on Swing in the Free World is ongoing so you'll probably need to do some CVS magic to get a version of gij that's capable of running it; I don't know whether any attempt has been made to get it into Kaffe yet. While you're at it, I also suggest apt-getting mono and downloading ikvm from www.ikvm.net. Run all your Java code on the.NET VM, with full inter-language interoperability. Keep your options open;) )
Any source for the claim that IBM has their own set of class libraries under development?
Seems to me that IBM would be stupid to do this when they could achieve the desired result much faster by cooperating with the existing Classpath or Kaffe projects.
My impression from IBM's recent behavior is that they're not stupid. Ergo I wouldn't expect them to be wasting time on an independent class library implementation.
Calling Kaffe "PersonalJava 1.1 compliant" is a massive understatement. It's pretty much completely 1.1 compliant with huge chunks of 1.2-1.4. It can run Tomcat and Eclipse (in fact, I can't think of a major open source Java application that it can't run, although there may be some. Do any major open source Java apps use Swing?).
Kaffe, Classpath and GCJ are making HUGE strides in compatibility; mentioning "1.1" to a Java developer is like mentioning Windows95 to a modern windows user, and it's just not true!
Slashdot Mirror
It's easy - in fact, my comment gives just about all the description you need. Obviously, as with any method, you can write pages and pages about the consequences for strategic voting etc, but my reading on the topic has only pointed out one disadvantage: you have to rank everyone you approve of equally. Which is sort of a natural consequence of choosing to keep the ballot simple and not having ranked ballots.
Basically, strategic voting in an Approval system is a matter of choosing where to "draw the line" in your own preferences to decide who to vote for or not. For example, if my true preferences are Nader > Kerry > Bush, I have to decide whether I'm going to vote for Nader+Kerry or just Nader. That choice depends on how far apart the candidates are in my opinion, and how likely I think they are to win.
The ability/need to make this decision does have some arguably negative consequences (especially compared to Concordet which can reasonably claim to be entirely immune to strategy), but IMHO nothing compared to the negative consequences of all the other options, including Concordet (which has the IMHO fatal flaw that the ballot and counting processes are both too hard for the average voter).
I disagree that this would turn a state with three electoral votes into a battleground.
Consider that in order to actually get all three votes, you'd need to have a very large proportion of the popular vote (depending on how you round, somewhere between 75% and 85%). Then bear in mind that if you have *any* hope of achieving that proportion, you can probably guarantee at least one, if not two, of the electoral votes with no campaigning at all. Now you're fighting over one or two votes again.
Contrast to a state with 25 electoral votes, where even though you have to persuade more voters, you only have to shift 4%-5% of the voters statewide to get the same one vote as you do from Vermont.
I'm sure it's easier to effect a shift of opinion in 4% of a very large population than in 85% of a small population.
But the main point is that if *all* the states are battlegrounds, the candidates have to have broad appeal and not just to those in a few large states with close to 50-50 polling.
Same reason that I support Approval voting: "broad appeal" is exactly the feature a president should have!
1) IRV isn't better than plurality, it's worse.
2) "Just like IRV" means that you require the whole thirty second soundbite explaining IRV to happen first, so you exceeded your thirty seconds that way. Actually I think IRV will fail based on that criterion too - thirty seconds is longer than any quote I've seen on the news from any of the presidential candidates, or on any other topic for that matter. I can't remember seeing anything as complex as IRV *ever* explained on the news.
3) Approval voting: "Just like today except you get to vote for as many candidates as you like". That's less than a *5* second soundbite. Why go to all the trouble of explaing IRV in the first place, when (a) it sucks and (b) approval is so much simpler to explain?
Better than eliminating the electoral college would be to require each state to allocate its electors proportionately instead of winner-take-all for the state.
That would completely eliminate the concept of a "battleground state" as it exists now, and "florida" situations in the future - there would never be a situation where a small increase in real votes could net you 21 electoral votes in one shot. Any recounts would be, at most, fighting over one electoral vote at a time instead of a whole state's worth, because the margin of error is never so large that it would cover more than that proportion of the state's voters.
I think this would probably have to be federally or constitutionally mandated, because individual states that apply it to only themselves instantly *dis*advantage themselves: where they might previously have gotten lots of attention from the candidates because 20+ electoral votes were up for grabs, the candidates would now concentrate on the states that *hadn't* implemented the change.
Concordet is clearly complex and despite having spent some time trying to understand it, I still don't fully grok it.
But how the heck is Approval complex? Anyone who's ever used a computer UI can have it explained in a single sentence "The ballot's a list of checkboxes instead of radio buttons". Even without that, it's pretty easy to say "Just check off as many names as you like", or "It's just like what we have now except you can vote for more than one person".
It's actually *simpler* than the current system, because even the morons who voted Buchanan/Gore in 2000 would have had their votes counted!
On the other hand, again having researched this a little bit, I still have no idea how you would construct a paper ballot for IRV that *I* could understand easily, let alone those people. IRV seems to practically *demand* electronic voting machines - and we all know what the state of the art in those is like!
You need to go a little further, I think.
You need to require that the licensee grants to *everyone*, not just the licensor, a right to use all their patents used in the same product as the licensed patent (I'd like to say "all patents owned by the same person" but that's probably impractical) under either the terms of the "PGPL" or under completely unrestricted terms.
This is more like the actual behavior of the GPL: As soon as you combine GPL code with something else into a single product, you're required to distribute the whole thing under the GPL. Similarly, this proposed license would say that as soon as you use this patent along with some other patented technique in the same product, you have to release the other patent under the same terms.
Now, if IBM and Novell could both be persuaded to license their entire patent pools under terms like these, we might be getting somewhere.
There's one big gotcha that I can see: As long as you only require the "grant-back" for patents owned by the licensee, there's a loophole where the licensee simply sells their patents to someone else prior to doing the licensing, and then gets revenue from them. On the other hand, if you require it for all patents, owned or not, then you have the problem of code that inadvertently infringes one of the zillions of bogus patents - logically, this should be permitted, but under the license it wouldn't be.
Question: Isn't SS1 effectively a multi-stage vehicle, with White Knight being the first stage?
The key insight in SS1 appears to me that you can make the throwaway stages also function as manned, landable aircraft. If you made a much bigger version of White Knight, Mounted under that a much bigger SS1, and attached to *that* a normal-sized SS1, wouldn't you effectively have a three-stage fully reusable orbit-ready vehicle? Essentially you launch the third stage from the top of the current SS1 hop.
Obviously I'm fully aware that the problems are nowhere near as simple as I'm implying. Quite apart from the fact that you can't simply scale up SS1 or WK and expect them to continue working, SS1 would probably not work very well as a carrier (that's why WK looks so different than SS1) and WK could certainly not feather in the way that SS1 does to get back from the separation point so probably all three vehicles would need to be radically different from the current WK and SS1. But I don't see any problem in principle.
Firefox supports a whitelist of sites that you can xpinstall from. This was added in the Preview Release, I believe. If you look in the release notes of that version, there should be more information on the whitelist and how to change its contents. Emptying the whitelist will effectively disable installing extensions.
What's astonishing to me is that *nobody* mentions that the only reason he's so far ahead of previous records is that in all previous seasons you were booted after 5 wins. Of *course* someone doing very well is going to beat records by a lot - anyone who did this well on any previous season would have left the show three weeks ago with a paltry 150K or so.
Sure, the guy's good. But there's no way to tell how much better he is than previous 5-day winners, because those previous winners never got the chance to show what they could do.
If I were single the answer would absolutely be "Hell yeah!".
;)
Since I'm married with a daughter and a son on the way, I'd have to say no - unless I could persuade them to come with me.
I'd love to emigrate to Mars. If the option becomes available within my lifetime, I *know* I'll try to persuade them. My wife won't go for it, but who knows how my kids will feel when they grow up... maybe they can persuade Mommy
"'A bunch of less cars on the road.'
:)
An english translation please?"
Seems perfectly correct to me. A Segway is certainly "less car" than, say, a Mini Cooper
I just tested a gnome-terminal that I happened to have open, and click-drag-to-select-menu-item worked fine. I don't have the latest GNOME, but that makes your statement suspect to me. Which version of GNOME were you using when you saw this behavior?
So use Konsole or gnome-terminal (okay, not gnome-terminal if you value the speed of your shell - they *really* need to apply the year-old patch to fix the scrolling speed - but I'm sure there are other alternatives as well) which has 'copy' and 'paste' in the menus...
In every program I use in Linux (specifically, Mozilla, rdesktop, and various GNOME stuff) Ctrl-C/Ctrl-V work perfectly, and so does the select/middleclick system, and neither interferes with each other.
I haven't used KDE in a long time but I understand that they introduced the same behavior with Qt3/KDE3.
Unless you're using really ancient software, pretty much everything will work in *either* mode, or you can do what I do and use a combination of both (choosing whether to bother pressing Ctrl-C to copy depending on whether you're going to need to highlight something at the destination).
I'm really curious to understand how so many people manage to still have a problem with this. Are you perhaps expecting that since "everybody knows that select copies on Linux", Ctrl-V will paste the thing that you last selected, instead of the last thing you Ctrl-C'd, and not testing it to verify this? Or just assuming that selecting something will overwrite your Ctrl-C buffer? I'd like to believe that people would actually test these things before posting Ask Slashdots about it, but you have to wonder...
Ok, by the time I post this 300 other people will have said the same thing, but how come this is filed under "apache"?
Yeah, as I said in the original post, I've written code (and a naive computer player, which I wrote to mitigate the difficulty of finding other people to play against) to play it using a grid rendering like my poor-quality ascii-art rendition, in HTML tables.
I still have the code but it's a horrible perl hack with dependencies on software I'm reluctant to release out of sheer shame, and I don't have a public server to host it on.
I keep intending to rewrite it under Mono or something, but never get around to it. The AI is really simple but still quite hard to beat. I suppose now I have to try your suggestion of playing it in a 5^4 grid instead of 4^4...
Not sure if you're being tongue-in-cheek here, but if you're not and have actually played 5^4 tictactoe, what's it like to play? Is there a noticeable advantage to getting the center square?
Do you have any software for playing it?
Stuart.
4D tictactoe (naughts and crosses, for brits) is a very difficult game to master. I've written an implementation with a fairly naive computer player that just weights each square based on the sum of a weighting of each line that goes through it, and I still have a really hard time beating the computer at it. I guess the computer's just much better at visualizing 4 dimensions than I am ;)
/. to let me render this... I'll show a 2^4 grid instead of the normal 4^4 one to save typing...)
If you're interested in the implementation, I'm afraid it's not publically available right now, but it's not that hard to write. The main insight is that your entire 4D gui can be done in straight HTML tables (let's see if I can get
|_|_| |_|_|
|_|_| |_|_|
|_|_| |_|_|
|_|_| |_|_|
Why 4^4 rather than 3^4? It turns out that 3^n for any n>2 has an easy strategy that allows the first player to always win. Proving that is left as an exercise for the reader.
You're still missing the point.
My proposed Open Source iTunes store would honor the DRM. Of course, being Open Source it's impossible to stop someone else from coming along and taking that DRM code out again, but how many average users would bother to recompile an application that came with their distribution, when for the most part it would let them do what they wanted?
Remember, for most people burning to a CD and ripping it (which Apple allows, so my proposed app would too) is much easier than downloading a patch and recompiling the app. And of the people who would go to this trouble, how many of them wouldn't just go and get Playfair off some P2P network if they were using the Apple client instead? The existence of an Open Source client that honors the DRM is no worse than the existence of Apple's client with Playfair out of the bag (which it is, and it ain't going back in).
The point is that even if the actual app we build honors the DRM, the player itself has to "break" it to get to the content to allow it to be played. So, paradoxically, we need the DRM to be broken before we can write an Open Source app that honors it.
So many people are missing one important reason why PlayFair is important.
Think "iTunes store for Linux".
Think Open Source iTunes store for Linux.
Think someone reverse-engineering the protocol for iTunes store and allowing music to be downloaded - AND PAID FOR - on a Linux computer using only Open Source software.
That would be a big deal. Apple's never going to do it. Playfair is necessary (but not sufficient) to make it happen.
Furthermore, it's probably the only way that I'd ever use iTunes store. I haven't decided for sure whether I'd consider using a binary proprietary iTunes release for Linux, but I certainly won't be using a Windows or Mac version.
The Open Source client could even enforce the DRM when used unmodified - I'd still use it, and I wouldn't hack the source to remove it. Probably some people would, but those people already have Playfair anyway. I have no need to bypass those restrictions, but I do need a Linux client if I'm going to use the thing at all.
So in a small way, Playfair would allow Apple to make a little more money (by selling to Linux users - a small but not non-existent, and growing market) without costing them a cent in development costs.
Red Hat, eh? I'm surprised you don't get "apt-get: command not found", but I guess times change...
:) A quick google doesn't reveal any good kaffe RPM sources - perhaps the Kaffe developers all use different distributions.
Heh, I doubt that adding debian unstable to your sources-list would help much - dpkg is probably not included with RH9
If you're comfortable building from source, the tarball from kaffe.org shouldn't cause any problems.
I heard that Red Hat is pushing gcj fairly heavily as a solution, so I would expect that gij is either present by default on RH9 or easy to install using whatever normal means Red Hat users use for getting software.
Hope this helps,
Stuart.
apt-get install kaffe gcj gij
.NET VM, with full inter-language interoperability. Keep your options open ;) )
kaffe com.whatever.MyApp
gij com.whatever.MyApp
HTH HAND
(Btw, Kaffe is merging its class libraries with GNU Classpath and is able to run many current apps, such as Tomcat and Eclipse. The version in debian unstable is (at long last) pretty current. I don't know what the debian status of gcj/gij is. Work on Swing in the Free World is ongoing so you'll probably need to do some CVS magic to get a version of gij that's capable of running it; I don't know whether any attempt has been made to get it into Kaffe yet. While you're at it, I also suggest apt-getting mono and downloading ikvm from www.ikvm.net. Run all your Java code on the
(goodbye, URL!)
Any source for the claim that IBM has their own set of class libraries under development?
Seems to me that IBM would be stupid to do this when they could achieve the desired result much faster by cooperating with the existing Classpath or Kaffe projects.
My impression from IBM's recent behavior is that they're not stupid. Ergo I wouldn't expect them to be wasting time on an independent class library implementation.
So, again, do you have a source for that claim?
Calling Kaffe "PersonalJava 1.1 compliant" is a massive understatement. It's pretty much completely 1.1 compliant with huge chunks of 1.2-1.4. It can run Tomcat and Eclipse (in fact, I can't think of a major open source Java application that it can't run, although there may be some. Do any major open source Java apps use Swing?).
Kaffe, Classpath and GCJ are making HUGE strides in compatibility; mentioning "1.1" to a Java developer is like mentioning Windows95 to a modern windows user, and it's just not true!