Note, another factor is that it's easier to integrate an in-kernel server with the exported filesystems.

File locking, NFSv4 file delegations, NFSv4 change attributes, and lookup of files by filehandle are all areas where we've traditionally depended on in-kernel interfaces between knfsd and filesystems.

Over time we've added APIs which expose that functionality to userspace (see for example open_by_handle_at(2)), and we may some day get to the point where userspace has access to all the same filesystem functionality as knfsd, but we're not there yet.

There actually is an active userspace NFS server project (Ganesha). My impression is that it's currently better suited for exporting filesystems (typically complicated distributed systems) that themselves run mainly in userspace.

First-semester physics and you can't quickly produce the basic equations for acceleration, momentum, kinetic energy? Learning to do symbolic differentiation and don't know the product rule? I think you likely have a problem.

Point taken that you don't necessarily have to sit down and memorize them (I never did), though I think there's no harm in doing that and it might help. I'd agree that not knowing them is more a symptom than a cause. But it's a potentially serious symptom: if you're approaching the exam and don't know the product rule, I'd rather the message be "you've got a problem and need a lot more practice", not "don't worry, you can look it up someplace".

But again the real problem with internet-connected devices in exams isn't their use as a reference but as a person-to-person communication device. The exam *does* need to test the student, not somebody else....

I've taught introductory calculus and we definitely didn't reuse exams.

There are some things you need to be able to have at your fingertips without having to google for them each time.

But even that aside, my big worry wouldn't be that they didn't memorize the quadratic formula or something, it'd be that they paid somebody to go sit on the other side of a chat session and coach them through the test. At that point we're really not testing the student any more.

Memorizing formulas is the easy part of learning anything, I don't know why that's the thing some students obsess about, and as a teacher I'd be concerned about students not memorizing things they probably should, but it's not the end of the world.

It's when they start trying to message someone to get help that I'd get really worried.... It's not likely to work as well as they think it will, but I still wouldn't want to have to deal with it.

We used them when I was teaching introductory calculus as a grad student in the 90's.

A smartphone's certainly capable enough, but I can still think of a number of advantages to a special-purpose calculator:

• If you're using it an hour at a time in class, something with dedicated calculator buttons is probably going to be more comfortable than the touchscreen interface.
• There's less to go wrong. In a class with 30 students, I'd be afraid one of them would always have a dead phone battery or a crash or.... (Or worse, I would while I'm trying to demonstrate something.)
• They're generally No wireless networking, so you can give test problems that might require calculators without having to deal with the whole "how do I know you aren't texting with someone during the test" problem.

But sure maybe some day it will make sense to require everyone to have a phone and standardize on some single calculator app.

Note T-Mobile does offer a $30 "unlimited" month-at-a-time plan (data up to 5 gigs, then throttled, unlimited texts, 100 minutes, if I remember right), and there are people who do the math and end up buying an unlocked GSM phone to go with that. I don't know how many, but it has an attraction beyond hard-core android geeks.

(I don't know whether that meets your requirements for a "decent data plan". It works OK for my purposes.)

Ingo is talking about application ABI, not driver ABI. He's objected to proprietary drivers before, and I don't recall seeing any evidence of a change of heart.

Thanks to virtual memory and so on, applications have much more of an arms-length relationship to the rest of the system than drivers. That makes maintaining a fixed driver ABI more work.

In the judgement of most kernel developers, the most efficient use of limited resources is to write an open source driver that can be included in the upstream kernel. That makes it easier for other kernel developers to collaborate with your your developers on future maintenance.

In that situation the open source driver writer may work for the hardware company, or sign an NDA with them, and then they can call the same guy.

Never done it, don't know the details (what do they do to ensure nothing's unintentionally disclosed in the source code?). But it seems to be a common arrangement.

"The Scrabble word that bothers me is "aa". I mean seriously."

Why is this bothering you? Are you getting a little tense? Maybe it's time for a vacation somewhere? Hawaii is nice....

English-speaking people live in enough different places and have sufficiently diverse cultures and interests that there will *always* be words that seem obviously common words to group X and obviously made up to group Y. If you need a hard-and-fast decision for something like a board game, then you have to choose some authority to make the word-by-word judgement call.

I've heard English speakers use "qi" in English sentences, but can't ever recall hearing anyone use "quoi" on its own in an English sentence, so until we get an ascii-32 tile I think Scrabble is safe from "je ne sais quoi".

Words are imported from other languages all the time, and it's a judgement call when to start calling them English words. For a game like Scrabble where you need a black-and-white decision in each case, that means the only way to have a complete set of rules is to agree on a dictionary. There will be inevitably be cases where someone could dispute a dictionary's choices.

What rule do you think would disallow "pi"? I have a hard time with that, especially on Slashdot, especially around this time of year. And people really do write it "pi", at least as often as the actual Greek character, and not just because they have trouble finding the latter on the their keyboards.

A few ways of looking at that $400k figure:

- Assuming 50k TSA employees (from wikipedia), that's $8 a year each. So in a good week they find two dimes.
- The TSA budget is about 8 billion. This is one two-thousandth of a percent of their budget.
- The 2012 budget is a few trillion. If congress passed a bill like Jeff Miller's every hour of every day that would cover... um, I think I have this right... one tenth of one percent of the budget. Looking to cut government waste? How about electing people who won't waste everyone's time on trivia.

In short: look, I'm not particularly fond of airport security either, but this is stupid.

In addition to those they give (VCR's used for time-shifting, image thumbnailing), I'd also add copying a program into RAM for the purposes of executing it (assuming ownership of the copy on disk). (Distressingly, my memory is that the status of this isn't clear in the US, but elsewhere in the world I think it is).

Again, that's circular reasoning, when the question we're trying to ask is exactly the question of whether people should have this particular right.

Some rights don't have to be granted: we take them to be, as they say, "self evident". I don't consider the right to have a government-granted monopoly on certain texts or ideas as among those inherent moral rights.

Note, that's not the same as saying that I don't think such rights should be granted--I think they *should* be--I'm only saying that we do it mainly for practical reasons (as an incentive to creation and dissemination, etc.) rather than a priori "moral" reasons.

But even if you do consider copyright a "moral" to some degree, I don't think that continues to a good guide once you get down to questions like "does the owner of a book need the author's permission to allow a 3rd party to scan the book for the sole purpose of searching it and presenting snippets giving context for the search results".

So it really does come down to the question of weighing the practical advantages and disadvantages to the world of granting that specific right.

And considered that way, the ability to "accomplish a dream" in one or the other case really is one of the factors to be weighed.

And full-text search of 10's of millions of books--that's a hell of a dream.

Fine. That's not the question, though--the question is *which uses* of copyrighted works should people have to pay for?

The Lord of the Rings movies required payment to whoever inherited Tolkien's rights.

The warehouses worth of epic fantasy written since then required none.

That's a lot of money people who got paid for stuff that wouldn't have existed without Tolkien having set pen to paper.

Most of us would argue that's a reasonable trade off: Tolkien didn't get everything he could be said to have deserved, but there was also a lot of interesting and creative work done that wouldn't have been if it all had to go through the bottleneck of his heirs.

That happens all the time. For-profit institutions have libraries too. The price I paid for one measly copy of K&R doesn't begin to approach the value of the income I've received from what I've learned from it.

Again, for me it comes down to weighing practical consequences: how much income are authors going to give up, versus what kinds of services are no longer going to be possible?

How?

A big university library has around 10 million books. Google claims a goal of scanning 130 million. There are something like a million more published a year.

Forget orphaned work or hard-to-reach authors; just consider authors that still individually hold rights to their books. How do you get legal permission from millions of them, at a per-book cost (I'm not even talking about the licensing; just the labor cost to contact them and verify that they agreed) that makes the whole project feasible?

And once you do, what do you actually expect to actually wring out of them for your 10 millionth (or whatever it works out to be) of the advertising income due to google books?

Hey, I'm just hand-waving here myself, but my intuition here is that if we require opt-in, we're just not going to get anything like Google books in our lifetimes, if ever.

"libraries have to buy copies of their books like everyone else, and they can still only lend one copy out to one person at a time. No extra copying or duplicated access is going on here"

So, the question is which uses of the copyrighted work require the rights-holders permissions, and which don't. It would be convenient to be able to draw a simple bright line at "copying", but that's never how it's worked:

- photocopying a small part of a book for your personal use in research is generally fine (but: go see reams of fair use guidelines for details).
- renting is complicated even though there's no copying involved (if I remember right: generally "first sale" permits it, but there are specific prohibitions in the US in the case of music).
- Using a character from a novel in a movie may involve no literal copying, but may require permission even when the plot and dialog is new.

Computers fuzz the lines too: you normally have to copy a program into RAM to run it, though most people would consider that ordinary use of a copy that they owned.

And personally I don't see a strong "moral" argument from one side or the other in these cases--I think the best way to decide where to draw these lines, however messy it may be, is weighing the likely practical consequences.