Or, rather, observing the degree of influence telcos have on regulators and legislators, Adam Smith would be likely to point out (from An Inquiry into the Nature And Causes of the Wealth of Nations, Book I, Chapter 11, Conclusion of the Chapter):
The interest of the dealers, however, in any particular branch of trade or manufactures, is always in some respects different from, and even opposite to, that of the public. To widen the market and to narrow the competition, is always the interest of the dealers. To widen the market may frequently be agreeable enough to the interest of the public; but to narrow the competition must always be against it, and can serve only to enable the dealers, by raising their profits above what they naturally would be, to levy, for their own benefit, an absurd tax upon the rest of their fellow-citizens. The proposal of any new law or regulation of commerce which comes from this order ought always to be listened to with great precaution, and ought never to be adopted till after having been long and carefully examined, not only with the most scrupulous, but with the most suspicious attention. It comes from an order of men whose interest is never exactly the same with that of the public, who have generally an interest to deceive and even to oppress the public, and who accordingly have, upon many occasions, both deceived and oppressed it.
I've used 14 languages and counting for development projects, including at least one of each of the interesting language types I know about (script, HDL, functional, pure imperative, assembly, compiled, bytecode compiled, purely graphical), . Ruby is among them. If you know another language or VM that gives IoC capability, why not share with the whole class?
It would probably be shorter to list the languages that don't give IoC capability; just about every modern language does that. Whether a particular framework uses it or not is a bigger question, though there are plenty of frameworks that do. The Wikipedia article on the type of IoC you are talking about that Spring provides for Java (dependency injection) lists 35 different frameworks (for ActionScript, C++, ColdFusion, Java,.NET, PHP4, PHP5, Perl, Python, and Ruby) providing that kind of IoC.
How old was OOP when it started being used for real development? 80 years old or so?
OOP isn't -- and digital computers aren't -- that old now. OOP, AFAIK, as a coherent paradigm emerged pretty much simultaneously to its use for practical development in the field of simulation in the 1960s.
A language like Perl or Python or Ruby, for example, is much higher-level than C. Writing a compiler in an HLL like this would be about a million times easier than writing it in C. You could target something like LLVM and still be portable.
There is something like that in each the Perl (with Perl6) and Ruby (Rubinius) worlds.
Of course it's also true that a spoon is next to useless for the purposes of building a skyscraper, but such is not the purpose of the spoon.
Well, yes, and the next time someone posts on a Slashdot discussion "Why don't they just build skyscrapers out of spoons", that'll be relevant. However, the question someone asked in this thread was (paraphrased) why don't people reverse-engineering.NET just use the MSDN documentation, so the answer that the MSDN documentation (however adequate its accuracy may be for other purposes) is not sufficiently accurate to support that is perfectly reasonable and appropriate, and your spoon-skyscraper comparison entirely misses the point of why the accuracy of MSDN was being discussed in the first place.
Last time I checked the biggest feeding frenzies are right after Thanksgiving and the weeks following Christmas. Of course sales are down to a trickle, everyone has cashed in their gift cards.
Sure, that is part of the explanation (as well as increasing bad news on the wider economic front) for the overall downturn in HD player sales (that is, HD DVD and Blu-Ray combined.)
OTOH, it doesn't explain why that lower combined total is comprised of a greater (absolute, not merely relative) number of Blu-Ray players and a vastly lower number of HD-DVD players. Blu-Ray becoming the preferred format does explain that. (Other explanations are, of course, possible.)
The local store has moved hardly any 360's, iPods, PS3's, TV, laptops, etc, etc either. Does that mean it's the death knell for all of those products too?
If might for some of them, if, e.g., the drop-off in 360s and PS3s was accompanied by a huge increase in Wiis, or the drop-off in iPods was accompanied by a huge increase in some other alternative brand of music player. The issue here isn't that the sales of one class of product of gone down, it is that within a class, the sales of one competing alternative have dropped sharply while the other competing alternative has increased.
Java's VM is not very friendly to Perl and similar languages. Java has a strict belief in strong typing that extends to compiled bytecode. Closures are possible to emulate, but hackish. And you can completely forget multiple inheritance, "duck" typing [wikipedia.org] ("IS A" versus "CAN" or "DOES"), mixins, almost all metaprogramming, runtime re-blessing of object classes, runtime modification of inheritance hierarchies (useful on rare occasions), or tons of other bits from Perl, Python, Ruby, and all the other "cool" kids.
And yet, implementations of plenty of dynamic languages (Groovy, Python, Ruby, Scheme, etc.) exist on the JVM, often with comparable (or better) performance on similar hardware to the dominant native implementations, where non-JVM implementations exist.
It actually surprises me every time I hear about someone hacking a new language to compile down to Java bytecode. After all, Java VM bytecode is so intimately tied to Java that it's rather straightforward to decompile it from bytecode to Java source and get good results.
In short, whoever hacked up the latest Foo-to-bytecode compiler has also written a Foo-to-Java source converter for free.
I'm not sure why this is makes making a JVM compiler so surprising; making a Foo-to-C (because C compilers are ubiquitous pretty much everywhere) source converter as a way of building a compiler for new language Foo is hardly unheard of; Java taking that role on the JVM doesn't seem all that surprising to me.
. Stickers are obvious no-ops, but spoilers and exhausts can be useful. It's just that most don't, because they're not designed for it. A spoiler that's actually designed for aerodynamic (rather than visual) effect will diminish lift (or in extreme cases, create drag, but those are huge), thus improving traction, useful for high-speed handling.
Wandering OT, but I think when you say "create drag" you really mean "create negative lift"; all spoilers (even ones not designed for anything but visual effect) increase drag somewhat, which doesn't improve traction. Reducing lift (and, a fortiori, creating net negative lift) does increase traction, increasing handling at speed, acceleration, attainable top speed (as less traction due to lift means less thrust that the wheels can deliver, so reducing lift means increasing the speed at which the thrust that can be delivered equals the overall drag.)
This argument loses a lot of it's luster when you consider certain aspects of pay. Don't count it as dollars paid out but as a percentage of productivity.
How is that different that what I suggested in GP, which was that the additional cost needed a compensating demonstrable additional benefit?
This is especially true when your overtime is directly linked to the profitability of a company.
Yes, if your overtime produces additional profitability for a company, that's an additional benefit to weigh against the additional cost. Of course, if you have an environment in which the overtime has no additional (direct dollar, at least: there are probably morale and retention costs that are harder to measure) cost, then a smaller benefit will justify the decision to make people work overtime. When, in addition to soft costs that may be hard to assess, there is a clear dollar cost to the decision to compel people to work overtime, that hard cost is going to need a hard benefit to support it. Maybe not initially, as practices will take time to be reviewed, but eventually.
The boss hardly has to account for overtime in his budget if the end result of having his employees working overtime is that he pays them what he has always paid them.
Sure he does, if the end result of him not having them work overtime is that he pays them less. If he has a choice, and the choice has a cost, he'll be expected (sooner or later) to justify that the cost had a corresponding benefit. The relation of the cost to what costs were incurred before he had any choice, or when the choice had a different cost profile, aren't going to be very relevant.
Your reply also displays the fact that you don't know much about Prolog, otherwise you wouldn't be making comments about re-binding variables or updating running code. A Prolog dynamic rule assertion is a piece of _Prolog source code_ of arbitrary complexity that is added at run-time to the database (this is an internal Prolog construct, not a RDBMS, although some implementations support using a RDBMS as a persistent rule store). Code written in Prolog obviously requires something capable of "understanding" Prolog to be useful, hence the fact that compiled Prologs typically include some sort of Prolog interpreter in either the compiled code itself or its run-time support system.
"Something that understands Prolog", sure, but that need not be a Prolog interpreter. A number of compiled languages can add bits built from of arbitrary source code at runtime using something that "understands" the language -- a compiler. Off the top of my head, Erlang, C#, Mozart/Oz, newer releases of Java, and some Common Lisp implementations each have some mechanism to do this. Its possible that no Prolog implementations do work this way, but nothing about the ability to add constructs built from arbitrary source code and runtime requires a language to use an interpreter rather than a compiler. It certainly requires either an interpreter or a compiler to be available to the running program, but a compiler compiled into a compiled program is certainly one way this can be (and for many languages has been) provided.
Note also that an interpreter which requires two or more passes of the source for _an entire software system_ before it can start doing its job (i.e. interpreting) would be more correctly described as a VM whose input is derived from an in-memory compiler, because even insignificant changes to any part of that source will require re-compiling all of it.
That seems to rest on a fairly narrow definition of what isn't an interpreter, but its not like there is a real concrete divide between interpretation and compilation; its certainly more of a continuum. But also, the argument you make here doesn't follow. That all of the source must be analyzed before any is executed doesn't mean that changes couldn't be analyzed incrementally if the right information was stored from the initial analysis. Note here that I'm not asserting this is practical or desirable for Eiffel.
That Linux is secure despite having a small market share has been the gospel around here for as long as I can remember
Its never been gospel. Its been one side in an ongoing debate forever (the other side has always argued that Linux (or Mac OS, or FreeBSD, or whatever else, depending on the particular context) would have as many exploits as Windows if it were as popular.) But even so, the "Linux is secure" side has not generally argued that exploits located are independent of popularity, but that Linux has specific features which make it more secure (posters will often list the features they see as most key) which would result in it having fewer exploits of any given level of severity even if it were as popular as Windows. Most posters (though perhaps not a narrow fringe) acknowledge that being less popular as Windows is at least one factor which reduces the scale of efforts to exploit Linux.
So when you start using the exact opposite argument with Vista and it suddenly gets modded to +5 with equally positive followups, it is the same slashdot "mob" at work.
Eh, its not uncommon to see posts with opposite arguments get modded +5 in the same thread, and addressing the same topic. Slashdot moderation isn't particular consistent, and, even if it were, its quite possible for two views that are fundamentally opposed to both be interesting and/or insightful (and/or funny, for that matter.)
Then it's perfectly valid to point out the logical inconsistancy
Well, no, there is only a logical inconsistency if the same people are making the argument, and the arguments are actually incompatible. As pointed out above, you've misrepresented the unanimity of opinion on Slashdot (which is the only basis for suggesting that it is the same people making the argument), and you've misrepresented the actual dominant argument (which is the only basis for suggesting the arguments are actually incompatible).
'We have to find a way to make the aspects of capitalism that serve wealthier people serve poorer people as well,' Mr. Gates will say in a speech at the World Economic Forum in Davos, Switzerland.
The aspects of capitalism that most serve wealthier people do so largely because they impose burdens on poorer people, a cost wealthier people are largely willing to let other people pay for their benefit. The regulation of so-called "free" markets is largely driven by exactly the kind of mercantile forces Adam Smith warned against allowing to dictate policy, with the same kind of negative effects to the general interest that he warned that allowing that would have. You can have an economic system that serves the interest of poorer people (i.e., everyone but the super-rich) better than the current system does, and it might even fairly be called a form of "capitalism" (it might, for instance, be a reasonable extension of Adam Smith's ideas, at least moreso than our current system is), but it certainly won't rely on the aspects of status quo capitalism that serve the wealthiest now, it will, of necessity, need to mitigate or outright eliminate those aspects.
True but Gates headed MS for very long, and I think this is the pot calling the kettle a money shark...
Bill Gates, while less involved in daily operations, still heads Microsoft. The only difference is that while Microsoft still works just as much for Gates as it always has, Gates works less for Microsoft.
This isn't necessarily the case, because some languages are specifically designed for an interpreted environment. One example is Prolog, which has the ability to modify its behaviour at run-time by asserting new rules into its internal database.
Um, so? Lots of compiled languages can "change their behavior" at runtime by rebinding variables, or even, in some cases, updating running code. Nothing about that behavior is inconsistent with a compiled implementation, and most prolog implementations that aren't toys (and many that are) are compilers.
The other side of the coin would be represented by Eiffel for example, which requires global code analysis for some of its design-by-contract features, and is therefore ill suited to running in an interpreted environment.
How does that make it ill-suited to running in an interpreted environment? It certainly would require a two-pass interpreter that seperates analysis from execution, but that's not an uncommon interpreter design.
There are many programs that perform a variety of unrelated tasks, and cannot therefore be described by a single rule consisting of sub-rules.
There is no such thing as a set of rules that cannot be described as a single rule consisting of sub-rules. If "Do X", "Do Y", and "Do Z" are rules; "Do X, Y, and Z" is a single rule, of which "Do X", "Do Y", and "Do Z" are sub-rules. (More relevantly in the case of real programs with multiple "unrelated" functions, if "When C(X) Do A(X)", "When C(Y) Do A(Y)", "When C(Z) Do A(Z)" are rules, then "For subrule in [X,Y,Z] When C(subrule) Do A(subrule) End" (intepretted either a sequential loop or a concurrent construct) is a single rule that incorporates each of the others as subrules. The actual content of the subrules is immaterial, any set of rules that can be applied can be composed into a single rule.
I have always thought that most of the OS config should be moved to a proper networked DB.
You mean like the Windows registry?
I would imagine that a heirarchical key/value db wasn't the "proper networked DB" that the GP had in mind.
And while its ubiquitous, its not all that well integrated.
It hard to beat hand editable files. It's just too convenient.
There's no reason an OS which integrated such a database couldn't expose configuration "files" as if they were normal files in the file system, and let them be editted as such. And bad edits (ones that weren't formatted properly) could simply fail to be committed back to the db (perhaps producing an error that an application or shell could catch and report back).
So a crappy database server with some Java tacked on somewhere? No thanks. Microsoft has the advantage the SQL Server is actually a good database. If I wanted to choose something open source, I'd just pick postgresql, as it's a hell of a lot closer to the Oracle/SQL Server class of products.
MySQL is probably easier to buy. And PostgreSQL is under a BSD license, which means that if you've got the money to throw at the necessary development and integration, you can take anything you like from it and incorporate it into something you own with minimal trouble, something you can't do with MySQL unless you own it.
What that means is is you want to use it in propietry apps you have to buy a license.
Or, alternatively, buy MySQL AB, though that's a little more expensive, but gives you more freedom if you plan to include some or all of MySQL, including its client libraries, in different proprietary applications, or to have freedom to choose the terms under which it is relicensed without future renegotiation.
I know enough to know that the "Theory of Evolution" doesn't warrant the "theory" moniker, because it's a hypothesis, yet to be proven.
Then you don't know the first thing about science, which is that no model in science is ever proven, it is either tested and disproven or tested and yet to be disproven. And those tests are conducted by generating predictions from the model and attempting to verify them.
(You also don't seem to understand that "evolution" isn't, properly speaking, a single theory [or hypothesis], it is a body of theory, an entire area of research, and that the current models in evolution are, indeed, theories that have made predictions, and those predictions have been examined and not disproven.)
I have never seen (or heard of) cases where a direct lineage from one species to another is proven.
Again, you don't understand science. Science doesn't prove explanations, it draws testable conclusions from models, tests them, and rejects the model if the test fails*. Since its always possible that a model has implications not previously understood, and since future tests built on those implications may yet fail, nothign is ever proven in science, a model is either rejected or accepted provisionally.
There are ideas floating around, based on the examples that are found, but there's no evidence that one species became another.
Actually, there is considerable evidence that new species have arisen from existing species, including directly observed speciation. Of course, the tests of hypotheses whose subject matter is not directly observable are somewhat indirect, but that doesn't change the process, or its validity.
It isn't reproducible, etc.
The tests of various elements of evolutionary theory are, in fact, "reproducible, etc." The understanding that a particular set of more modern species evolved from a particular common ancestor species in the distant past may be a conjecture, a hypothesis, or a theory (there are all kinds of predictions that are testable that can be drawn and tested based on such a belief, and often they have been; some, however, have indications in the evidence but no tests proposed or made, and so may only be conjecture.)
But you make a category error if you confuse the fact that some specific widely-seen-as-likely chains of descent are not tested with the idea that evolutionary theory is itself mere speculation.
* this is a bit of a simplification; in fact, models which are disproven by testing generally continue to be dominant with the caveat that they fail in certain conditions for which there is no explanation until a new model is developed which explains the results for which the old model fails and the results for which it succeeded, then the new model becomes the dominant model.
After that, remember that believing in Evolution requires that one be worse at math than one thinks he is, and it also requires that one have more faith than the much-maligned Christian.
I'm a Christian, too, and that's just plain wrong. Believing in evolution (either in the sense of "believing it is true", or "believing it is the best model yet proposed for the way the world works in the area it covers", the latter being the usual sense of "belief" when discussing scientific theories, which is rather different from religious belief) requires neither of those things.
For open source fans, I think Parrot is our best bet for a VM to give.NET a fight (although feel free to reply with other suggestions, I don't keep up with too many others;-).
I can understand not following all the myriad VM options out there, but overlooking Java is a bit surprising. Already supports quite a lot of languages, already widely installed, supports calls into and out of dynamic languages to other code on the VM.
You can criticize Ruby for a lot of things, but to claim that it's uglier than Python is laughable.
Ruby retains some of Perl's potentially ugly features, though there are for most uses alternatives that are more idiomatic in current Ruby usage, so Ruby certainly can be ugly.
I think you're misunderstanding what I mean by I/O.
Since you were responding to what I said about how I/O was presented in different circumstances, I think the problem is that you were misunderstanding what I mean by I/O.
My use of the term is a generic one which does not imply any form of I/O devices, but refers instead to the fact that all meaningful processes (i.e. ones that do more than consume CPU time with empty loops) operate on data which can be described as an input to the process itself, and a result which is the process's output.
Which wasn't what I was talking about first courses delaying, at all, and I can't see how you could misread that, since the specific example was outputting "Hello, World" via System.out.println().
This is such a fundamental concept that any course which doesn't teach students to think of data and processes in this way from the outset is failing to do it's job.
Of course it is.
Agreed. I've said elsewhere in this topic that I favour interpreted languages as vehicles for teaching very basic concepts precisely because they have little or no overhead
Whether a language is "interpreted" has little to do with it, in fact, "interpreted" isn't a feature of languages, but of implementations. Lisps (often compiled) has as little overhead as Ruby (interpreted at least prior to 1.9; whether bytecode run on a VM is "interpreted" or "compiled"—or both—is a side topic, though it relates to which side of that fence Java should be considered on, as well.)
I fail to see where this is semantically different from my definition in any way that doesn't make it less generally applicable (e.g. "processes" can clearly include an arbitrary number of rules, whereas "a rule" implies that programs can only have one of them).
Using the broad use of I/O that you have no clarified that you were using (which makes your original criticism misplaced with regard to the post it responded to), there is little difference. (Its not less generally applicable than yours, since a rule need not be simple; a compound rule composed of other rules can still be described as "a rule".)
Assuming that you're asking because you're curious, don't know. Technical schools may be an example?
I was curious, and the first article (the one to which this is a "followup" of sorts) presented the problem as near-universal in American colleges and universities, and those schools that I thought to check to confirm it not only didn't use Java exclusively in their CS curricula, they didn't use it exclusively (or in many cases at all) in the first programming course intended for CS majors.
I'm kind of curious if there isn't a lot of churn being generated over a problem that doesn't really exist in the first place.
Slashdot Mirror
Or, rather, observing the degree of influence telcos have on regulators and legislators, Adam Smith would be likely to point out (from An Inquiry into the Nature And Causes of the Wealth of Nations, Book I, Chapter 11, Conclusion of the Chapter):
It would probably be shorter to list the languages that don't give IoC capability; just about every modern language does that. Whether a particular framework uses it or not is a bigger question, though there are plenty of frameworks that do. The Wikipedia article on the type of IoC you are talking about that Spring provides for Java (dependency injection) lists 35 different frameworks (for ActionScript, C++, ColdFusion, Java,
OOP isn't -- and digital computers aren't -- that old now. OOP, AFAIK, as a coherent paradigm emerged pretty much simultaneously to its use for practical development in the field of simulation in the 1960s.
Its almost exactly the opposite: its more of Rails, repackaged to appeal to enterprise users.
What makes you say that Groovy is a "limited, powerless, unexpressive language"?
There is something like that in each the Perl (with Perl6) and Ruby (Rubinius) worlds.
Well, yes, and the next time someone posts on a Slashdot discussion "Why don't they just build skyscrapers out of spoons", that'll be relevant. However, the question someone asked in this thread was (paraphrased) why don't people reverse-engineering
Sure, that is part of the explanation (as well as increasing bad news on the wider economic front) for the overall downturn in HD player sales (that is, HD DVD and Blu-Ray combined.)
OTOH, it doesn't explain why that lower combined total is comprised of a greater (absolute, not merely relative) number of Blu-Ray players and a vastly lower number of HD-DVD players. Blu-Ray becoming the preferred format does explain that. (Other explanations are, of course, possible.)
If might for some of them, if, e.g., the drop-off in 360s and PS3s was accompanied by a huge increase in Wiis, or the drop-off in iPods was accompanied by a huge increase in some other alternative brand of music player. The issue here isn't that the sales of one class of product of gone down, it is that within a class, the sales of one competing alternative have dropped sharply while the other competing alternative has increased.
And yet, implementations of plenty of dynamic languages (Groovy, Python, Ruby, Scheme, etc.) exist on the JVM, often with comparable (or better) performance on similar hardware to the dominant native implementations, where non-JVM implementations exist.
I'm not sure why this is makes making a JVM compiler so surprising; making a Foo-to-C (because C compilers are ubiquitous pretty much everywhere) source converter as a way of building a compiler for new language Foo is hardly unheard of; Java taking that role on the JVM doesn't seem all that surprising to me.
Wandering OT, but I think when you say "create drag" you really mean "create negative lift"; all spoilers (even ones not designed for anything but visual effect) increase drag somewhat, which doesn't improve traction. Reducing lift (and, a fortiori, creating net negative lift) does increase traction, increasing handling at speed, acceleration, attainable top speed (as less traction due to lift means less thrust that the wheels can deliver, so reducing lift means increasing the speed at which the thrust that can be delivered equals the overall drag.)
How is that different that what I suggested in GP, which was that the additional cost needed a compensating demonstrable additional benefit?
Yes, if your overtime produces additional profitability for a company, that's an additional benefit to weigh against the additional cost. Of course, if you have an environment in which the overtime has no additional (direct dollar, at least: there are probably morale and retention costs that are harder to measure) cost, then a smaller benefit will justify the decision to make people work overtime. When, in addition to soft costs that may be hard to assess, there is a clear dollar cost to the decision to compel people to work overtime, that hard cost is going to need a hard benefit to support it. Maybe not initially, as practices will take time to be reviewed, but eventually.
Sure he does, if the end result of him not having them work overtime is that he pays them less. If he has a choice, and the choice has a cost, he'll be expected (sooner or later) to justify that the cost had a corresponding benefit. The relation of the cost to what costs were incurred before he had any choice, or when the choice had a different cost profile, aren't going to be very relevant.
"Something that understands Prolog", sure, but that need not be a Prolog interpreter. A number of compiled languages can add bits built from of arbitrary source code at runtime using something that "understands" the language -- a compiler. Off the top of my head, Erlang, C#, Mozart/Oz, newer releases of Java, and some Common Lisp implementations each have some mechanism to do this. Its possible that no Prolog implementations do work this way, but nothing about the ability to add constructs built from arbitrary source code and runtime requires a language to use an interpreter rather than a compiler. It certainly requires either an interpreter or a compiler to be available to the running program, but a compiler compiled into a compiled program is certainly one way this can be (and for many languages has been) provided.
That seems to rest on a fairly narrow definition of what isn't an interpreter, but its not like there is a real concrete divide between interpretation and compilation; its certainly more of a continuum. But also, the argument you make here doesn't follow. That all of the source must be analyzed before any is executed doesn't mean that changes couldn't be analyzed incrementally if the right information was stored from the initial analysis. Note here that I'm not asserting this is practical or desirable for Eiffel.
Its never been gospel. Its been one side in an ongoing debate forever (the other side has always argued that Linux (or Mac OS, or FreeBSD, or whatever else, depending on the particular context) would have as many exploits as Windows if it were as popular.) But even so, the "Linux is secure" side has not generally argued that exploits located are independent of popularity, but that Linux has specific features which make it more secure (posters will often list the features they see as most key) which would result in it having fewer exploits of any given level of severity even if it were as popular as Windows. Most posters (though perhaps not a narrow fringe) acknowledge that being less popular as Windows is at least one factor which reduces the scale of efforts to exploit Linux.
Eh, its not uncommon to see posts with opposite arguments get modded +5 in the same thread, and addressing the same topic. Slashdot moderation isn't particular consistent, and, even if it were, its quite possible for two views that are fundamentally opposed to both be interesting and/or insightful (and/or funny, for that matter.)
Well, no, there is only a logical inconsistency if the same people are making the argument, and the arguments are actually incompatible. As pointed out above, you've misrepresented the unanimity of opinion on Slashdot (which is the only basis for suggesting that it is the same people making the argument), and you've misrepresented the actual dominant argument (which is the only basis for suggesting the arguments are actually incompatible).
The aspects of capitalism that most serve wealthier people do so largely because they impose burdens on poorer people, a cost wealthier people are largely willing to let other people pay for their benefit. The regulation of so-called "free" markets is largely driven by exactly the kind of mercantile forces Adam Smith warned against allowing to dictate policy, with the same kind of negative effects to the general interest that he warned that allowing that would have. You can have an economic system that serves the interest of poorer people (i.e., everyone but the super-rich) better than the current system does, and it might even fairly be called a form of "capitalism" (it might, for instance, be a reasonable extension of Adam Smith's ideas, at least moreso than our current system is), but it certainly won't rely on the aspects of status quo capitalism that serve the wealthiest now, it will, of necessity, need to mitigate or outright eliminate those aspects.
Bill Gates, while less involved in daily operations, still heads Microsoft. The only difference is that while Microsoft still works just as much for Gates as it always has, Gates works less for Microsoft.
Um, so? Lots of compiled languages can "change their behavior" at runtime by rebinding variables, or even, in some cases, updating running code. Nothing about that behavior is inconsistent with a compiled implementation, and most prolog implementations that aren't toys (and many that are) are compilers.
How does that make it ill-suited to running in an interpreted environment? It certainly would require a two-pass interpreter that seperates analysis from execution, but that's not an uncommon interpreter design.
There is no such thing as a set of rules that cannot be described as a single rule consisting of sub-rules. If "Do X", "Do Y", and "Do Z" are rules; "Do X, Y, and Z" is a single rule, of which "Do X", "Do Y", and "Do Z" are sub-rules. (More relevantly in the case of real programs with multiple "unrelated" functions, if "When C(X) Do A(X)", "When C(Y) Do A(Y)", "When C(Z) Do A(Z)" are rules, then "For subrule in [X,Y,Z] When C(subrule) Do A(subrule) End" (intepretted either a sequential loop or a concurrent construct) is a single rule that incorporates each of the others as subrules. The actual content of the subrules is immaterial, any set of rules that can be applied can be composed into a single rule.
I would imagine that a heirarchical key/value db wasn't the "proper networked DB" that the GP had in mind.
And while its ubiquitous, its not all that well integrated.
There's no reason an OS which integrated such a database couldn't expose configuration "files" as if they were normal files in the file system, and let them be editted as such. And bad edits (ones that weren't formatted properly) could simply fail to be committed back to the db (perhaps producing an error that an application or shell could catch and report back).
MySQL is probably easier to buy. And PostgreSQL is under a BSD license, which means that if you've got the money to throw at the necessary development and integration, you can take anything you like from it and incorporate it into something you own with minimal trouble, something you can't do with MySQL unless you own it.
Or, alternatively, buy MySQL AB, though that's a little more expensive, but gives you more freedom if you plan to include some or all of MySQL, including its client libraries, in different proprietary applications, or to have freedom to choose the terms under which it is relicensed without future renegotiation.
Then you don't know the first thing about science, which is that no model in science is ever proven, it is either tested and disproven or tested and yet to be disproven. And those tests are conducted by generating predictions from the model and attempting to verify them.
(You also don't seem to understand that "evolution" isn't, properly speaking, a single theory [or hypothesis], it is a body of theory, an entire area of research, and that the current models in evolution are, indeed, theories that have made predictions, and those predictions have been examined and not disproven.)
Again, you don't understand science. Science doesn't prove explanations, it draws testable conclusions from models, tests them, and rejects the model if the test fails*. Since its always possible that a model has implications not previously understood, and since future tests built on those implications may yet fail, nothign is ever proven in science, a model is either rejected or accepted provisionally.
Actually, there is considerable evidence that new species have arisen from existing species, including directly observed speciation. Of course, the tests of hypotheses whose subject matter is not directly observable are somewhat indirect, but that doesn't change the process, or its validity.
The tests of various elements of evolutionary theory are, in fact, "reproducible, etc." The understanding that a particular set of more modern species evolved from a particular common ancestor species in the distant past may be a conjecture, a hypothesis, or a theory (there are all kinds of predictions that are testable that can be drawn and tested based on such a belief, and often they have been; some, however, have indications in the evidence but no tests proposed or made, and so may only be conjecture.)
But you make a category error if you confuse the fact that some specific widely-seen-as-likely chains of descent are not tested with the idea that evolutionary theory is itself mere speculation.
* this is a bit of a simplification; in fact, models which are disproven by testing generally continue to be dominant with the caveat that they fail in certain conditions for which there is no explanation until a new model is developed which explains the results for which the old model fails and the results for which it succeeded, then the new model becomes the dominant model.
I'm a Christian, too, and that's just plain wrong. Believing in evolution (either in the sense of "believing it is true", or "believing it is the best model yet proposed for the way the world works in the area it covers", the latter being the usual sense of "belief" when discussing scientific theories, which is rather different from religious belief) requires neither of those things.
I can understand not following all the myriad VM options out there, but overlooking Java is a bit surprising. Already supports quite a lot of languages, already widely installed, supports calls into and out of dynamic languages to other code on the VM.
If by "break" you mean "render standards-compliant pages correctly where the current IE fails to", that would seem to be correct.
I'm not sure that's a reasonable definition of "break".
Ruby retains some of Perl's potentially ugly features, though there are for most uses alternatives that are more idiomatic in current Ruby usage, so Ruby certainly can be ugly.
Ruby can also be not-ugly.
Since you were responding to what I said about how I/O was presented in different circumstances, I think the problem is that you were misunderstanding what I mean by I/O.
Which wasn't what I was talking about first courses delaying, at all, and I can't see how you could misread that, since the specific example was outputting "Hello, World" via System.out.println().
Of course it is.
Whether a language is "interpreted" has little to do with it, in fact, "interpreted" isn't a feature of languages, but of implementations. Lisps (often compiled) has as little overhead as Ruby (interpreted at least prior to 1.9; whether bytecode run on a VM is "interpreted" or "compiled"—or both—is a side topic, though it relates to which side of that fence Java should be considered on, as well.)
Using the broad use of I/O that you have no clarified that you were using (which makes your original criticism misplaced with regard to the post it responded to), there is little difference. (Its not less generally applicable than yours, since a rule need not be simple; a compound rule composed of other rules can still be described as "a rule".)
I was curious, and the first article (the one to which this is a "followup" of sorts) presented the problem as near-universal in American colleges and universities, and those schools that I thought to check to confirm it not only didn't use Java exclusively in their CS curricula, they didn't use it exclusively (or in many cases at all) in the first programming course intended for CS majors.
I'm kind of curious if there isn't a lot of churn being generated over a problem that doesn't really exist in the first place.