What we don't like is the current welfare system that does not encourage people to get off the welfare system. The current system is broken and broken badly.
Remember "the end of welfare as we know it"? Clinton signed the bill that forces people off the rolls after a couple years, and also requires working crap jobs to get anything at all. Does that "encourage people to get off the system" to you?
We *could* actually provide enough of a safety net to actually help people move up out of poverty, rather than just enough to prevent them from starving outright. Take away the work requirement, let it run long to wrap up a high school diploma and get a start on college, and pay enough for a home of some sort, day care for kids, and transportation to make it to a job.
I develop using OpenJDK and deploy on IBM JRE. I've yet to encounter any incompatibility between them. Are you actually aware of any proprietary extensions in the IBM Java runtime?
I used to run the IBM JRE a lot under Linux, and at the time there were some IBM-isms that emerged:
1) The JCE (crypto) API in the IBM JRE was a completely different implementation from Sun's. In practice this meant that exceptions had a different stack trace.
2) The font support under Linux/AIX was dramatically different. The IBM JRE often required custom command-line properties to work correctly on DBCS systems (if it worked at all).
3) The JIT behaved differently. In practice this often meant IBM's was much faster. But it also caused different kinds of segfaults than Sun's JRE.
These are the ones I remember. This was back in the 1.2.x - 1.3.x days. In those days I could find one reproducible JVM bug (often segfault) per month, but different ones for Sun vs IBM JREs.
I was transferred from WebSphere to the Tivoli building in RTP (Bldg 610) basically right after the Friday beer bash was gone - circa late 2002. I think maybe a couple people still did the beer thing, but certainly not most of the building. By that time most of the old Tivoli-ers were gone, and many of the IBMers who remained were pretty much the people who didn't have a better offer elsewhere. (When I finally got RA'd myself, several others from 610 were in my career counseling group. We all pretty much agreed that none of the products we worked on in 610 made sense from a business standpoint to even exist.)
The building was pretty looking though, and had a nicer cafeteria than WebSphere's.
You really think something as complicated as programming will be something the majority of the population will know?
Is writing absurdly easier than programming? I don't mean being capable of writing down a grocery list, I mean being able to write an informative essay or a narrative story. You may think it is, but then if your education was similar to mine you've had 12-15 years of writing practice in school and college, from forming letter shapes to spelling words and diagramming sentences on up to writing 20 page papers. If you had had a decade of programming classes starting at age 6 you would find programming as easy as you find reading or writing to be now. (When I went through public school, there was a total of six weeks of learning computer programming out of the entire 12-year curriculum. Maybe there is more programming in the curriculum now.)
Farming is far more vital to day to day life, yet we have few farmers. Some peopel specialize at it and make enough food for everyone else.
They are only capable of that because farming is a tangible good where economies of scale can work well. Software isn't producing a tangible good, it's putting a thought process down in concrete form, just like writing.
Contemporary example: At every white-collar job I have worked, everyone has to deal with Microsoft Excel. Accountants, shipping clerks, administrative assistants, receptionists, engineers, doctors, managers, marketers, project managers, B2B liasons, everyone uses Excel. And every time someone opens up Excel, they are entering a realm where programming might make their job easier. Yet almost every one of those Excel sheets is unique to that situation. There is no way a few dozen or even a few hundred "Excel sheet companies" could provide all of the Excel sheets that everyone else uses in their jobs. Worse, we have obsoleted the ability to do those jobs without something like Excel: we can't go back to paper balance sheets, hand-crafted Gantt charts, etc. In our relentless drive for growth, we have caused Excel to become the new basic literacy. I believe that once enough people are used to programming in their day jobs, it too will become the new basic literacy.
I know that geeks think that being able to program a computer is the be-all, end-all of being smart but it really isn't.
I personally don't think programming is the ultimate bar for anything. I just think economics will force it to become more fundamental to modern life. I used to be paid to program full time, but I left the software industry in 2004 to be a chemical engineer. From a programming perspective, my skills are a bit dated (I've never touched Ajax or cloud computing or Plone/Facebook/Rails/etc.). But my non-bleeding-edge Java and Visual Basic skills have proven very valuable at work in my engineering role -- I've been able to perform certain optimizations that other engineers struggle with just because I know how to put together some I/O and simple calculations. One of those just saved the company $250K this year; the next might be worth $120K. My company seems to see me as an engineer-who-can-program, not a programmer-who-can-engineer. They also have accountants-who-can-program who produce more bang for the salary buck than accountants-who-can't. Same with chemists.
I actually wish modernity would slow down a bit and leave room for people to NOT have to be so business-skilled just to survive. I wish we could have more artists, musicians, amateur sports players, etc., who could make a decent living.
If users mainly want the "familiar experience", then just give them the familiar experience - it's that simple - there is no good reason why a Linux-based Netbook, in 2009, can't have a similar interface and layout for most of the main functionality to Windows XP, which is now so ancient there has been plenty of time to create window managers that are clones to the pixel (or "similar but better").
I'm actually hoping ReactOS gets there. Basically it can be "Windows done right", i.e. Windows look and feel + Unixy features (virtual desktops, fully-supported GNU command line) + Windows driver support + proper security defaults + fully-supported protocols.
That would be an awesome first step for anyone wanting to jailbreak the Microsoft ecosystem. ReactOS clients + Linux servers could bring a lot of power to many people.
Has anyone entertained the thought that people might actually choose windows because it's the best available option?
I do all the time. There is no way in the world I would teach Linux to my parents or siblings, because for the most part they are rote learners with respect to computers. (The downside is that they periodically need the old wipe-it-clean method of virus/spyware removal every few years.) And for Java development, I actually prefer running Eclipse under Windows than under Linux, but I can't quite say why. Maybe I just don't have a fast enough Linux box for it.
The solution to so many things seems to be is looking at source code, or messing with a script and so on.
The corresponding problem on the Windows/Apple side is that many times there is NO solution to a problem, you just have to live with it.
I think sometimes OSS people forget that most of the people in the world are not programmers. Not just a majority, but like 99.999% or more.
I think in the distant future programming will be a skill so fundamental that everyone will be taught it like arithmetic. The same proportion of people who won't be able to program in the future will probably mirror the proportion of people who cannot do basic arithmetic today.
So you gave up on Linux after trying only one distro several years ago? You should give it another try, Linux has gotten a lot better since then.
I was talking to a recent hire engineer who was exposed to Linux circa 2003 in a college mathematics lab. He thought all Linux sucked because USB sticks and floppies didn't automount. He didn't know that those Linux boxes were most likely replacing Sun workstations.
What the math department perceived was "wow, Linux is so much nicer than the Suns we used to have." What he perceived was "Linux sucks because it doesn't behave like Windows".
If you want an objective certification, you need objective criteria. "mysidia calls project X successful" doesn't cut it. Market acceptance depends on luck, and usability is in the eye of the beholder.
And 5+ years industry experience is more than "just working". I didn't mention it so maybe you didn't know, but for those five years one has to be mentored by other licensed engineers. For example, if you're hired by Dow Chemical and spend five years doing accounting, that won't cut it for a Chemical Engineering PE license. You'd need to do five years of actual project work such as process improvements (cost savings), operations support (solving mysteries), design, training, etc. You can't take the final PE exam until some other PE signs off that you are ready for it.
For software, I can't see a practical way to objectively determine reasonable competency. I know where you're coming from with the point that just working in IT doesn't make you a decent programmer. But then I'm not sure enough people can agree on what DOES make a good programmer. We all know that novice programmers can make some truly awful code, but then again one can have greatly superior code-fu and still be a net drain on a development team if no one else can keep up with them. Yet we don't want to grant professional licenses to the mediocre middle.
Personally, I see programming as little more than refined arithmetic. It's a skill that everyone should have, right next to reading and writing. Future schools could call it the 4 R's if they want (Reading, wRiting, aRithmetic, pRogramming). How do we certify high-quality arithmeticians? Or quality readers or writers?
but also require real-world-like things to finalize or complete the cert (such as doing a project that was actually successful)
A certification can't have "actually successful" as a criteria because success is not dependent on technical skill but rather luck of right-product-at-the-right-time. And in fact popular products are often technically crap -- would you want the "genius" behind MySpace to get certification but prevent it for the contractor(s) that made the Mars rover successful?
Real engineering certifications (i.e. Professional Engineering licenses) have a technical bar (degree in related discipline + EIT/FE exam) and a practicum bar (5+ years relevant industrial experience). A software engineering certification should be similar. However, setting that kind of bar would require legislative action to establish a binding certifying authority, and I don't see the modern practitioners of professional IT (including myself) accepting such oversight.
You've got a paradox: who is more likely to screw up the vote? The election officials, or a person of authority outside the polling station? If the former, then the voter needs something they can take out with them to ensure their vote was cast the right way. If the latter, then the voter can't have anything leave the booth that could reveal their vote to someone else.
Which one do you think in practice is the greater threat to democracy?
A voting receipt wouldn't have the actual vote, it would have instead an identification number that the voter could use to look up their vote on a web site. In the event of a recount, all you need is a statistically random sample (yeah yeah hard to get blah blah) of voters to check that the web site has their votes accurately recorded and you'd have good confidence that the final tally is OK.
(such as the right to license under non-open-source licenses). As a consequence, the project isn't run as an open source project.
Nice to see that by your definition no GPL project is really open source, since only the original authors / copyright holders can turn GPL code into proprietary code.
I heard from many people who had become IBMers through acquisition (mainly Tivoli and the company behind the first WebSphere). Suffice to say when the "traditional IBM" started taking over things started to suck.
It would take some pretty nice benefits to get me to go back.
There was a time I thought they'd buy Sun just to own Java, but now that its been open sourced, that reasoning is out the window.
Sun still has the Java(tm) trademark. If they say something (e.g. Apache Harmony) isn't Java-compliant, then it's not. Having the ability to control the Java brand might be worth more money to IBM than the Java codebase itself.
It's possible that IBM is wanting to buy Sun to keep a customer. Back in the early 2000's, Informix was in trouble, so Walmart basically begged IBM to buy Informix and so that they would have support for their purchasing database (at the time the largest SQL database in the world). I'm sure that someone out there relies on Sun and is big enough to convince IBM that it would be worth it.
I suppose it depends on whether any of the vm86 mode in Windows 9x made it into Win2k and beyond.
If any of the 16-bit ASM code behind the various int 21h DOS calls was retained in the real-mode emulation layer then one could say modern Windows still has DOS code still in it.
Sure it can! All that has to happen is for the GPL code copyright holders to be contacted and given a reasonable opportunity to object to the change.
No. When you contact a copyright holder, it isn't "I will change the license unless you say no," it's "Can I please change the license?" If the copyright holder says no, you're done.
In the case of the Linux kernel, some of the copyright holders are dead, at least one is in prison (ahem), and huge chunks of critical code is copyrighted by corporations that have zero incentive to give Microsoft such a huge advantage.
A huge multi-year effort to push the kernel to GPLv3 (or even GPL v2 or later) might feasibly result in enough of a critical mass of code to change; after a few years of replacing the components that are still GPLv2 only the new codebase might become the de facto standard. But there is absolutely no chance in hell of enough code in the kernel moving to a non-GPL'ish license to make a proprietary fork.
Java truly is becoming the COBOL of today: a stagnant, no longer actively developed language in which tons of existing code are written and needs to be maintained. This isn't meant to be disparaging, and there's certainly a place for such languages;
The same could have been said of Fortran, C, and C++ at different times, but all of those have had revivals and new language features added. Even COBOL had several major updates over its lifetime.
Right now Java-the-language is near a pretty decent place for "enterprise" type development. It's easy to put a dozen people on one project and watch milestones get completed. The common language extensions are pretty good: annotations, AspectJ, and JUnit go a long way in quality control, and generics achieve the 80% benefit of C++ templates without the headaches of Turing-completeness. I think pushing the language much further would be more painful than just breaking out of the OOP paradigm and going with a dynamic language on the JVM like Clojure. Then you end up with a consistent platform in the JVM, Java as its assembly language, and all the syntactical/grammatical goodness you need for fuzzy problems. But the last thing Java needs is to become C++ without pointers.
but if you want reasonably modern language design, Java isn't going to offer that to you, while C# still can.
I'm not familiar enough with C# to make any conclusions about it. If you are, my question would be, "What can C# deliver that Java + Clojure can't?" Bear in mind also that OS independence is a must for me: some of my code does occasionally run on supercomputers.
I used to code Java in the JDK 1.1 - 1.4 days. It sucked ass. It was slow, had weird dependencies on X11, required a lot of boilerplate code (such that up to half of the LOC could be logging inside exception blocks), had various JRE incompatibilities all over the place (such that some applications just couldn't be run bug-free on ALL of AIX/Solaris/Windows/Linux/Mac), and the reference JDK/JRE was Sun's proprietary property. I left Java and went on to C, Perl, C++, and Lisp. Naturally I used Emacs and SLIME.
Then I found Clojure. And I got a $350 laptop from Walmart last week that had 3 gigs of RAM and a single-core 1.8GHz AMD processor. And I thought, "I wonder if Eclipse will run decent on this thing?" And it does, and it's not all that slow, and it is by far the best IDE I've ever used.
I'm now re-climbing the learning curve on modern Java, and it's looking pretty good now. AspectJ does a good job eliminating a lot of repetitive code, eclipse-metrics warns me when I'm not being decent at OOP design, and the available libraries are top-notch. Java the language isn't so bad anymore, and now with Clojure on top I have plenty of linguistic room to prototype and get to choose the best among many paradigms for each situation.
Slashdot Mirror
Note to AC: do not reply to yourself with a "note to self" using your real ID.
What we don't like is the current welfare system that does not encourage people to get off the welfare system. The current system is broken and broken badly.
Remember "the end of welfare as we know it"? Clinton signed the bill that forces people off the rolls after a couple years, and also requires working crap jobs to get anything at all. Does that "encourage people to get off the system" to you?
We *could* actually provide enough of a safety net to actually help people move up out of poverty, rather than just enough to prevent them from starving outright. Take away the work requirement, let it run long to wrap up a high school diploma and get a start on college, and pay enough for a home of some sort, day care for kids, and transportation to make it to a job.
I develop using OpenJDK and deploy on IBM JRE. I've yet to encounter any incompatibility between them. Are you actually aware of any proprietary extensions in the IBM Java runtime?
I used to run the IBM JRE a lot under Linux, and at the time there were some IBM-isms that emerged:
1) The JCE (crypto) API in the IBM JRE was a completely different implementation from Sun's. In practice this meant that exceptions had a different stack trace.
2) The font support under Linux/AIX was dramatically different. The IBM JRE often required custom command-line properties to work correctly on DBCS systems (if it worked at all).
3) The JIT behaved differently. In practice this often meant IBM's was much faster. But it also caused different kinds of segfaults than Sun's JRE.
These are the ones I remember. This was back in the 1.2.x - 1.3.x days. In those days I could find one reproducible JVM bug (often segfault) per month, but different ones for Sun vs IBM JREs.
Your ideas are intriguing to me and I wish to subscribe to your newsletter.
I was transferred from WebSphere to the Tivoli building in RTP (Bldg 610) basically right after the Friday beer bash was gone - circa late 2002. I think maybe a couple people still did the beer thing, but certainly not most of the building. By that time most of the old Tivoli-ers were gone, and many of the IBMers who remained were pretty much the people who didn't have a better offer elsewhere. (When I finally got RA'd myself, several others from 610 were in my career counseling group. We all pretty much agreed that none of the products we worked on in 610 made sense from a business standpoint to even exist.)
The building was pretty looking though, and had a nicer cafeteria than WebSphere's.
Para. 1: "... alcohol is anathema to business - in customer facing roles ..."
Para. 2: "... essential lubricant - high level sales, conventions, customers who offer it to salesmen."
You don't think we have a fundamental inconsistency here? I'm not poking fun at the poster, but at business mores.
Simple: one set of customers contains rich people, the other does not. Rich people can drink under any circumstance.
You really think something as complicated as programming will be something the majority of the population will know?
Is writing absurdly easier than programming? I don't mean being capable of writing down a grocery list, I mean being able to write an informative essay or a narrative story. You may think it is, but then if your education was similar to mine you've had 12-15 years of writing practice in school and college, from forming letter shapes to spelling words and diagramming sentences on up to writing 20 page papers. If you had had a decade of programming classes starting at age 6 you would find programming as easy as you find reading or writing to be now. (When I went through public school, there was a total of six weeks of learning computer programming out of the entire 12-year curriculum. Maybe there is more programming in the curriculum now.)
Farming is far more vital to day to day life, yet we have few farmers. Some peopel specialize at it and make enough food for everyone else.
They are only capable of that because farming is a tangible good where economies of scale can work well. Software isn't producing a tangible good, it's putting a thought process down in concrete form, just like writing.
Contemporary example: At every white-collar job I have worked, everyone has to deal with Microsoft Excel. Accountants, shipping clerks, administrative assistants, receptionists, engineers, doctors, managers, marketers, project managers, B2B liasons, everyone uses Excel. And every time someone opens up Excel, they are entering a realm where programming might make their job easier. Yet almost every one of those Excel sheets is unique to that situation. There is no way a few dozen or even a few hundred "Excel sheet companies" could provide all of the Excel sheets that everyone else uses in their jobs. Worse, we have obsoleted the ability to do those jobs without something like Excel: we can't go back to paper balance sheets, hand-crafted Gantt charts, etc. In our relentless drive for growth, we have caused Excel to become the new basic literacy. I believe that once enough people are used to programming in their day jobs, it too will become the new basic literacy.
I know that geeks think that being able to program a computer is the be-all, end-all of being smart but it really isn't.
I personally don't think programming is the ultimate bar for anything. I just think economics will force it to become more fundamental to modern life. I used to be paid to program full time, but I left the software industry in 2004 to be a chemical engineer. From a programming perspective, my skills are a bit dated (I've never touched Ajax or cloud computing or Plone/Facebook/Rails/etc.). But my non-bleeding-edge Java and Visual Basic skills have proven very valuable at work in my engineering role -- I've been able to perform certain optimizations that other engineers struggle with just because I know how to put together some I/O and simple calculations. One of those just saved the company $250K this year; the next might be worth $120K. My company seems to see me as an engineer-who-can-program, not a programmer-who-can-engineer. They also have accountants-who-can-program who produce more bang for the salary buck than accountants-who-can't. Same with chemists.
I actually wish modernity would slow down a bit and leave room for people to NOT have to be so business-skilled just to survive. I wish we could have more artists, musicians, amateur sports players, etc., who could make a decent living.
If users mainly want the "familiar experience", then just give them the familiar experience - it's that simple - there is no good reason why a Linux-based Netbook, in 2009, can't have a similar interface and layout for most of the main functionality to Windows XP, which is now so ancient there has been plenty of time to create window managers that are clones to the pixel (or "similar but better").
I'm actually hoping ReactOS gets there. Basically it can be "Windows done right", i.e. Windows look and feel + Unixy features (virtual desktops, fully-supported GNU command line) + Windows driver support + proper security defaults + fully-supported protocols.
That would be an awesome first step for anyone wanting to jailbreak the Microsoft ecosystem. ReactOS clients + Linux servers could bring a lot of power to many people.
Has anyone entertained the thought that people might actually choose windows because it's the best available option?
I do all the time. There is no way in the world I would teach Linux to my parents or siblings, because for the most part they are rote learners with respect to computers. (The downside is that they periodically need the old wipe-it-clean method of virus/spyware removal every few years.) And for Java development, I actually prefer running Eclipse under Windows than under Linux, but I can't quite say why. Maybe I just don't have a fast enough Linux box for it.
The solution to so many things seems to be is looking at source code, or messing with a script and so on.
The corresponding problem on the Windows/Apple side is that many times there is NO solution to a problem, you just have to live with it.
I think sometimes OSS people forget that most of the people in the world are not programmers. Not just a majority, but like 99.999% or more.
I think in the distant future programming will be a skill so fundamental that everyone will be taught it like arithmetic. The same proportion of people who won't be able to program in the future will probably mirror the proportion of people who cannot do basic arithmetic today.
So you gave up on Linux after trying only one distro several years ago? You should give it another try, Linux has gotten a lot better since then.
I was talking to a recent hire engineer who was exposed to Linux circa 2003 in a college mathematics lab. He thought all Linux sucked because USB sticks and floppies didn't automount. He didn't know that those Linux boxes were most likely replacing Sun workstations.
What the math department perceived was "wow, Linux is so much nicer than the Suns we used to have." What he perceived was "Linux sucks because it doesn't behave like Windows".
MySQL use is bigger than you try to make us believe.
The GP was talking technical merit ("awesome"), not popularity.
(seriously, I love PostgreSQL), but it is not even close to Oracle (DB2, maybe)
And there you go showing your ignorance. DB2 has been a better database than Oracle for quite a while.
If you want an objective certification, you need objective criteria. "mysidia calls project X successful" doesn't cut it. Market acceptance depends on luck, and usability is in the eye of the beholder.
And 5+ years industry experience is more than "just working". I didn't mention it so maybe you didn't know, but for those five years one has to be mentored by other licensed engineers. For example, if you're hired by Dow Chemical and spend five years doing accounting, that won't cut it for a Chemical Engineering PE license. You'd need to do five years of actual project work such as process improvements (cost savings), operations support (solving mysteries), design, training, etc. You can't take the final PE exam until some other PE signs off that you are ready for it.
For software, I can't see a practical way to objectively determine reasonable competency. I know where you're coming from with the point that just working in IT doesn't make you a decent programmer. But then I'm not sure enough people can agree on what DOES make a good programmer. We all know that novice programmers can make some truly awful code, but then again one can have greatly superior code-fu and still be a net drain on a development team if no one else can keep up with them. Yet we don't want to grant professional licenses to the mediocre middle.
Personally, I see programming as little more than refined arithmetic. It's a skill that everyone should have, right next to reading and writing. Future schools could call it the 4 R's if they want (Reading, wRiting, aRithmetic, pRogramming). How do we certify high-quality arithmeticians? Or quality readers or writers?
but also require real-world-like things to finalize or complete the cert (such as doing a project that was actually successful)
A certification can't have "actually successful" as a criteria because success is not dependent on technical skill but rather luck of right-product-at-the-right-time. And in fact popular products are often technically crap -- would you want the "genius" behind MySpace to get certification but prevent it for the contractor(s) that made the Mars rover successful?
Real engineering certifications (i.e. Professional Engineering licenses) have a technical bar (degree in related discipline + EIT/FE exam) and a practicum bar (5+ years relevant industrial experience). A software engineering certification should be similar. However, setting that kind of bar would require legislative action to establish a binding certifying authority, and I don't see the modern practitioners of professional IT (including myself) accepting such oversight.
You've got a paradox: who is more likely to screw up the vote? The election officials, or a person of authority outside the polling station? If the former, then the voter needs something they can take out with them to ensure their vote was cast the right way. If the latter, then the voter can't have anything leave the booth that could reveal their vote to someone else.
Which one do you think in practice is the greater threat to democracy?
A voting receipt wouldn't have the actual vote, it would have instead an identification number that the voter could use to look up their vote on a web site. In the event of a recount, all you need is a statistically random sample (yeah yeah hard to get blah blah) of voters to check that the web site has their votes accurately recorded and you'd have good confidence that the final tally is OK.
(such as the right to license under non-open-source licenses). As a consequence, the project isn't run as an open source project.
Nice to see that by your definition no GPL project is really open source, since only the original authors / copyright holders can turn GPL code into proprietary code.
"Me too."
I heard from many people who had become IBMers through acquisition (mainly Tivoli and the company behind the first WebSphere). Suffice to say when the "traditional IBM" started taking over things started to suck.
It would take some pretty nice benefits to get me to go back.
There was a time I thought they'd buy Sun just to own Java, but now that its been open sourced, that reasoning is out the window.
Sun still has the Java(tm) trademark. If they say something (e.g. Apache Harmony) isn't Java-compliant, then it's not. Having the ability to control the Java brand might be worth more money to IBM than the Java codebase itself.
It's possible that IBM is wanting to buy Sun to keep a customer. Back in the early 2000's, Informix was in trouble, so Walmart basically begged IBM to buy Informix and so that they would have support for their purchasing database (at the time the largest SQL database in the world). I'm sure that someone out there relies on Sun and is big enough to convince IBM that it would be worth it.
I suppose it depends on whether any of the vm86 mode in Windows 9x made it into Win2k and beyond.
If any of the 16-bit ASM code behind the various int 21h DOS calls was retained in the real-mode emulation layer then one could say modern Windows still has DOS code still in it.
Sure it can! All that has to happen is for the GPL code copyright holders to be contacted and given a reasonable opportunity to object to the change.
No. When you contact a copyright holder, it isn't "I will change the license unless you say no," it's "Can I please change the license?" If the copyright holder says no, you're done.
In the case of the Linux kernel, some of the copyright holders are dead, at least one is in prison (ahem), and huge chunks of critical code is copyrighted by corporations that have zero incentive to give Microsoft such a huge advantage.
A huge multi-year effort to push the kernel to GPLv3 (or even GPL v2 or later) might feasibly result in enough of a critical mass of code to change; after a few years of replacing the components that are still GPLv2 only the new codebase might become the de facto standard. But there is absolutely no chance in hell of enough code in the kernel moving to a non-GPL'ish license to make a proprietary fork.
Java truly is becoming the COBOL of today: a stagnant, no longer actively developed language in which tons of existing code are written and needs to be maintained. This isn't meant to be disparaging, and there's certainly a place for such languages;
The same could have been said of Fortran, C, and C++ at different times, but all of those have had revivals and new language features added. Even COBOL had several major updates over its lifetime.
Right now Java-the-language is near a pretty decent place for "enterprise" type development. It's easy to put a dozen people on one project and watch milestones get completed. The common language extensions are pretty good: annotations, AspectJ, and JUnit go a long way in quality control, and generics achieve the 80% benefit of C++ templates without the headaches of Turing-completeness. I think pushing the language much further would be more painful than just breaking out of the OOP paradigm and going with a dynamic language on the JVM like Clojure. Then you end up with a consistent platform in the JVM, Java as its assembly language, and all the syntactical/grammatical goodness you need for fuzzy problems. But the last thing Java needs is to become C++ without pointers.
but if you want reasonably modern language design, Java isn't going to offer that to you, while C# still can.
I'm not familiar enough with C# to make any conclusions about it. If you are, my question would be, "What can C# deliver that Java + Clojure can't?" Bear in mind also that OS independence is a must for me: some of my code does occasionally run on supercomputers.
I used to code Java in the JDK 1.1 - 1.4 days. It sucked ass. It was slow, had weird dependencies on X11, required a lot of boilerplate code (such that up to half of the LOC could be logging inside exception blocks), had various JRE incompatibilities all over the place (such that some applications just couldn't be run bug-free on ALL of AIX/Solaris/Windows/Linux/Mac), and the reference JDK/JRE was Sun's proprietary property. I left Java and went on to C, Perl, C++, and Lisp. Naturally I used Emacs and SLIME.
Then I found Clojure. And I got a $350 laptop from Walmart last week that had 3 gigs of RAM and a single-core 1.8GHz AMD processor. And I thought, "I wonder if Eclipse will run decent on this thing?" And it does, and it's not all that slow, and it is by far the best IDE I've ever used.
I'm now re-climbing the learning curve on modern Java, and it's looking pretty good now. AspectJ does a good job eliminating a lot of repetitive code, eclipse-metrics warns me when I'm not being decent at OOP design, and the available libraries are top-notch. Java the language isn't so bad anymore, and now with Clojure on top I have plenty of linguistic room to prototype and get to choose the best among many paradigms for each situation.
Give Java a fresh look, it's come a long way.