Yes, Dr Statistician, I know you're not a programmer, but that thing you're writing is a program, and you will use revision control or you are not working on my project.
I think it's great that R is written by statisticians for statisticians, and that statisticians find it useful. But you shouldn't then be surprised if it doesn't do parallelism, even though many statistical problems on large data sets should intuitively be easy to do in a parallel kind of way.
Of course Python doesn't beat R; it doesn't even compete in the same space. (Oh, and its formal semantics are almost as bad as that of R.)
Of course Julia doesn't beat R. It was designed to be a modern SISAL (which in turn was designed to be a modern Fortran). Again, it doesn't pretend do the same thing as R.
The thing is, the computer science community knows it can do better than R. The problem is convincing the statisticians that a better R is in their best interests.
When it comes to highly poisonous snakes, I think scared (or at least cautious) is a perfectly rational response.
The reason why Australian animals are considered scary is that every intuition that you have about what constitutes "dangerous" is completely and utterly wrong.
Australia has three of the five most venomous (which, incidentally, is different from "poisonous"; a poisonous animal is only a problem if you try to eat it) snakes in the world. The number of deaths from these snakes in most years since the development of antivenin is exactly zero, and moreover it only seems to bites herpetologists. If you don't want to get bitten, change jobs.
Tasmanian devils are truly scary, but their chief defence mechanism when confronted by humans is to stay as still as possible and shiver nervously. The worst experience that you're going to get from them is a bad night's sleep if they get under your house.
As for spiders, a good rule of thumb is that the bigger and hairier it is, the less dangerous it is. Even then, there are more deaths per annum from allergic reactions to bees.
So what Australian animals do you need to be scared of?
Well, if you stay out of Queensland (especially the northern part), you'll never see a box jellyfish, stonefish, or cassowary. You'll definitely want to avoid cassowaries, by the way. They are living proof that dinosaurs never became extinct. They look like an overgrown turkey, but they can and will beat the living crap out of you. However, there is only one confirmed human death by cassowary, and it was a teenager who decided it might be good idea to beat one with a club. The cassowary disagreed.
The blue ringed octopus is pretty nasty, but stay out of the water in the wrong time of year and you won't encounter one.
But the scariest of them all, to my mind, is the platypus. Yes, you head that right. I have two words for you: mammalian venom.
Platypus venom won't kill you, but you'll wish that it had. We have no antivenin treatment for it, so you're in for a a couple of months of an experience where the phrase "a world of pain" doesn't really do it justice.
You also have to see the situation from the point of view of Galileo's scientific critics. He was known to love a drink. Scratch that, he was known to love a lot of drink. He designed his improved contraption to look at the night sky, and reported little moving lights around Jupiter.
His fellow scientists had good reason to be very skeptical about his claim. We might have had to wait another hundred years if it hadn't been for Kepler.
Are you by any chance referring to Bruno, whose "hypothesis" was a mystical vision? Or are you referring to Copernicus, who actually had calculations and was not excommunicated?
You don't need total confiscation. When you need to crack down on citizens, all you need is [A] that they don't own handguns (because those are primarily defensive weapons), and [B] that all other weapons are registered.
It's easier to just convince gun owners that whoever is the target of tyranny is their enemy too. This is how the US government got away rounding up US citizens and putting them in internment camps.
Oh, also convince them that "we" respect your rights (as we're collecting all your phone calls) and "they" are trying to take your guns away (even though they're not). That shit always works.
You raise a good point, which is that capital punishment is only one of the aspects of the US crime industry which is completely messed up. Elected judges, elected prosecutors, plea bargains, and grand juries are other anachronisms which most countries have never had, or have abolished because they are not considered just.
I fail to understand why "throw away the key" is not serving "justice". Who is this "justice", and why does he or she demand human sacrifice? Are you afraid that the crops will be spoiled if the evildoer is not removed?
The higher death penalty rate for African Americans results from a much higher rate at which African Americans commit the kinds of crimes that receive the death penalty.
Educated white sociopaths tend to become CEOs, who can kill hundreds or thousands at a time and get off scot free.
I don't know anyone (sensible) who wants GC or reflection.
Nobody wants "reflection" exactly, but lots of people want some kind of meta-object protocol. Any time you need a scripting layer, you need that. Just imagine how much nicer Qt would be without moc.
But you generally only want it when you want it. Unfortunately, you pay for a MOP even when you don't use it.
My implication in posting that quote was "if D was popular, people would be complaining about it too", because all languages have a determined set of detractors (anti-Java "not everything fits into OO", anti-Python "whitespace isn't a substitute for program structure", anti-Lisp "how many brackets do you need")....
I'm glad you mentioned that, because I got the exact converse. If D was popular, more people would be trying to do very hard things in D, and so the problems in the language would be more apparent.
The best critics of C++ are hard-core C++ users, because their criticisms are usually accurate. People who complain about bloat have, almost all of the time, never used C++ for anything nontrivial. I know this partly because I used to say the same things; I didn't understand why the C++ language isn't bloated until I learned from bitter experience why all that is there. (The standard libraries, on the other hand, are definitely bloated.)
The third rule is usually quoted as "don't optimise until you've profiled".
But I would add two important rules:
1. If you must optimise something while coding, optimise your internal APIs and your automated tests. That way, when you a performance problem, you can swap code out, swap new code in, and everything should just work.
2. The best time and place to do optimisation is often before you write code, on the back of an envelope.
Lens flare can be a fine artistic choice, but needs to be applied with a very light touch. The Praxis explosion from the start of ST6 and the torch scene from Toy Story are two great examples of how lens flare can make a shot. A little goes a long way.
Slashdot Mirror
If you're a truly great programmer, then you will refuse to write any code in certain languages.
Yes, Dr Statistician, I know you're not a programmer, but that thing you're writing is a program, and you will use revision control or you are not working on my project.
Woah, sorry, had a flashback there.
Ah, but from the point of view of a computer scientist, the "best tool for the job" isn't necessarily the best tool that currently exists. R is a fabulous set of well-documented algorithms and linked together with one of the bizarre, poorly-specified and inadequately-documented language with a flaky, abstraction-leaking, poorly-performing implementation.
I think it's great that R is written by statisticians for statisticians, and that statisticians find it useful. But you shouldn't then be surprised if it doesn't do parallelism, even though many statistical problems on large data sets should intuitively be easy to do in a parallel kind of way.
Of course Python doesn't beat R; it doesn't even compete in the same space. (Oh, and its formal semantics are almost as bad as that of R.)
Of course Julia doesn't beat R. It was designed to be a modern SISAL (which in turn was designed to be a modern Fortran). Again, it doesn't pretend do the same thing as R.
The thing is, the computer science community knows it can do better than R. The problem is convincing the statisticians that a better R is in their best interests.
The reason why Australian animals are considered scary is that every intuition that you have about what constitutes "dangerous" is completely and utterly wrong.
Australia has three of the five most venomous (which, incidentally, is different from "poisonous"; a poisonous animal is only a problem if you try to eat it) snakes in the world. The number of deaths from these snakes in most years since the development of antivenin is exactly zero, and moreover it only seems to bites herpetologists. If you don't want to get bitten, change jobs.
Tasmanian devils are truly scary, but their chief defence mechanism when confronted by humans is to stay as still as possible and shiver nervously. The worst experience that you're going to get from them is a bad night's sleep if they get under your house.
As for spiders, a good rule of thumb is that the bigger and hairier it is, the less dangerous it is. Even then, there are more deaths per annum from allergic reactions to bees.
So what Australian animals do you need to be scared of?
Well, if you stay out of Queensland (especially the northern part), you'll never see a box jellyfish, stonefish, or cassowary. You'll definitely want to avoid cassowaries, by the way. They are living proof that dinosaurs never became extinct. They look like an overgrown turkey, but they can and will beat the living crap out of you. However, there is only one confirmed human death by cassowary, and it was a teenager who decided it might be good idea to beat one with a club. The cassowary disagreed.
The blue ringed octopus is pretty nasty, but stay out of the water in the wrong time of year and you won't encounter one.
But the scariest of them all, to my mind, is the platypus. Yes, you head that right. I have two words for you: mammalian venom.
Platypus venom won't kill you, but you'll wish that it had. We have no antivenin treatment for it, so you're in for a a couple of months of an experience where the phrase "a world of pain" doesn't really do it justice.
So now you know: don't fuck with a platypus.
There's no mystery here. Republicans and Democrats are both in favour of kill lists, drone strikes, and lobbyist corruption.
You're right that this isn't necessarily the case, but people back in the day knew that people saw things when they were drunk.
You also have to see the situation from the point of view of Galileo's scientific critics. He was known to love a drink. Scratch that, he was known to love a lot of drink. He designed his improved contraption to look at the night sky, and reported little moving lights around Jupiter.
His fellow scientists had good reason to be very skeptical about his claim. We might have had to wait another hundred years if it hadn't been for Kepler.
One of the few Slashdot threads where discussion of tentacle-porn biomechanoid double penetration is completely on topic.
Are you by any chance referring to Bruno, whose "hypothesis" was a mystical vision? Or are you referring to Copernicus, who actually had calculations and was not excommunicated?
Pics or it won't happen.
So where are all of those NRA protests against wholesale warrantless wiretapping?
It's easier to just convince gun owners that whoever is the target of tyranny is their enemy too. This is how the US government got away rounding up US citizens and putting them in internment camps.
Oh, also convince them that "we" respect your rights (as we're collecting all your phone calls) and "they" are trying to take your guns away (even though they're not). That shit always works.
Yeah, its only justification for existing is that it would probably be more effort to disallow it.
You raise a good point, which is that capital punishment is only one of the aspects of the US crime industry which is completely messed up. Elected judges, elected prosecutors, plea bargains, and grand juries are other anachronisms which most countries have never had, or have abolished because they are not considered just.
I fail to understand why "throw away the key" is not serving "justice". Who is this "justice", and why does he or she demand human sacrifice? Are you afraid that the crops will be spoiled if the evildoer is not removed?
Educated white sociopaths tend to become CEOs, who can kill hundreds or thousands at a time and get off scot free.
Since nobody in their right mind would start a new C++ application without Boost these days, there's always Boost.Build. It's pretty nice.
Nobody wants "reflection" exactly, but lots of people want some kind of meta-object protocol. Any time you need a scripting layer, you need that. Just imagine how much nicer Qt would be without moc.
But you generally only want it when you want it. Unfortunately, you pay for a MOP even when you don't use it.
C++, Objective-C, and SQL are the most obvious ones. ML and Haskell (while not as "major") also took a long time to find their feet.
I'm glad you mentioned that, because I got the exact converse. If D was popular, more people would be trying to do very hard things in D, and so the problems in the language would be more apparent.
The best critics of C++ are hard-core C++ users, because their criticisms are usually accurate. People who complain about bloat have, almost all of the time, never used C++ for anything nontrivial. I know this partly because I used to say the same things; I didn't understand why the C++ language isn't bloated until I learned from bitter experience why all that is there. (The standard libraries, on the other hand, are definitely bloated.)
My two favourite languages are C++ and Haskell (not in that order), so I feel for you, dude.
The third rule is usually quoted as "don't optimise until you've profiled".
But I would add two important rules:
1. If you must optimise something while coding, optimise your internal APIs and your automated tests. That way, when you a performance problem, you can swap code out, swap new code in, and everything should just work.
2. The best time and place to do optimisation is often before you write code, on the back of an envelope.
The syntax being built on top of C is pretty dated.
(I love C++, but I think I really want to program in D.)
The standard library implementation of locales.
Oh, you meant the language, not the library? Protected inheritance, then.
Yeah, that's not good.
Lens flare can be a fine artistic choice, but needs to be applied with a very light touch. The Praxis explosion from the start of ST6 and the torch scene from Toy Story are two great examples of how lens flare can make a shot. A little goes a long way.