Which is why it makes an easy target for a Scheme compiler, right?

http://github.com/dyoo/moby-scheme
http://planet.plt-scheme.org/display.ss?package=moby.plt&owner=dyoo
http://www.cs.brown.edu/~sk/Publications/Talks/Moby-Bootstrap/

Actually, imperative languages are typically based on procedures, not functions. The fact that such languages tend to call procedures "functions" is the confusing bit.

xmonad does what you want with desktops my default. Each of my monitors is bound to one specific workspace at a time; I can switch either monitor independently to any workspace, or manually stretch windows across the gap in between.

You can use xmonad as the window manager for GNOME.

It'll require a little bit of tweaking to make it look normal, though (you'll need to add window decorations, and configure it to make windows floating by default), or you could learn the keyboard shortcuts and use tiling, which sounds like it may work better with the way you want to think about screens anyway. (It sounds terrible, I know, but it's remarkably effective.)

Sure. It's all in what you consider "fundamental", I guess.

" but inefficient isn't the end of the world. Just means it's a first step towards something better."

That's like saying a break in AES to 2^40 isn't the end of the world, it just makes cracking it slightly more efficient. :-P

This was nothing fundamentally new; google "secure multiparty computation." Or, FTFA, Gentry's technique requires a "trillion times" more computational power than existing techniques.

Not that I think his work wasn't awesome-- I've already queued the paper in my reading list. All I'm claiming is that he didn't "solve a major open problem".

That's turboing.

FAT32: no POSIX metadata, 4GB file size limit is deadly. It's inefficient, and generally outdated and nasty.

NTFS: proprietary, sometimes complicated to get on Linux, hard to get on OSX.

ext*: bad-to-none support on non-Linux. IIRC, neither the Windows nor OSX drivers support journaling, for example.

HFS+: about the same boat as ext*, if you swap "Linux" and "OSX".

UDF: reasonably efficient, support for basic metadata (POSIX, though no EAs or forks), full support on Linux 2.6, OSX 10.5, Windows Vista/7, or (with third-party utils)

What you fail to understand is that the customer-accessible part of the manager food chain in the vast majority of companies is approximately two people tall: CSR and supervisor. (Depending on the company and nature of question, you may be able to get to tier2 support; hence the "approximately".)

You will have better luck...

• Just calling back. Virgin Mobile's policies used to differ depending on which call center your call got routed to, but even in less extreme cases, some reps are nicer than others.
• Turboing. In particular, some companies have started to have "Executive Support" hotlines (Sprint comes to mind.) Save these for a last resort. GetHuman is also useful.
• Moving horizontally. Try web order support, activations, billing, customer service, terminations, etc.
• Being nice instead of nasty.
• Writing. Yes, seriously. I've resolved many issues just by sending the entity in question a nastygram. People still take snail-mail seriously.

Maybe because it's decent in suburban areas, because it's significantly cheaper than AT&T, Sprint, or Verizon, because we don't mind roaming, and because T-Mobile is slightly less obviously run by assholes than the other three aforementioned companies?

Let me elaborate a bit on a previous comment.

T-Mobile's ETF is $200. They can't offer you a subsidy of significantly more than $200, because then it would be to your advantage to buy a phone, cancel your contract, and sell the phone for a profit. Contract lengths are two years. Amortized over 2 years, by taking the subsidy, you save $8._33. Meanwhile, T-Mobile's has two tiers of plans. One gives you the subsidy, has a contract, and is $10 more a month. Thus: if you have the money up-front, buy your phone, save $240 in the cost of a plan over two years, and save yourself from a contract.

Consider the Motorola Cliq. As a user, it's a good phone.

but they haven't even bothered to make sure that only leaf certificates can be issued.

Nope; the CA only signed a what you call a leaf certificate, but the constraint which determines whether a key can is a branch ("CA = true") or leaf ("CA = false") was part of the cert that they were able to change. See the last paragraph of section 5.1