What GP is saying is just plain wrong. If he were correct, aftermarket HIDs would be legal in the UK, and they're not.

This is what car manufacturers claim, but it's nowhere near true. The typical slewing time for the optics looks to be about 0.5 seconds from "standard" to "dipped", and that's way too slow for reacting to bumps/potholes when you're going faster than 30. I regularly get blinded by Audis, BMWs and Volvos with factory HIDs, but Mitsubishi seems to be worst. It could perhaps be that they are just worst at controlling chromatic abberations (since blue light blinds you more).

I sincerely doubt that is true. In Norway (which is not in the EU but strives to comply with EU regulations as far as cars are concerned) the limit has been in lumens for almost ten years. It's commonly called "E-marking" which is lumens and directionality combined and dumbed down for the regular person. Ordinary (non-HID) high beams are typically 12.5 points each, good external pencil beams are 40 points each, and you're at most allowed to have 100 points active simultaneously.

There is no way these could be "better" than existing solutions, at least not for high beams. There are stringent restrictions on how much light you are allowed to emit in which directions, and if you've bought some $400 Hella Rallye Compact's you're touching that limit no matter how bad your stock headlights are. The only usefulness here is that BMW will perhaps be able to make more exotic looking headlights.

Well, maybe because it's the strictest checking compiler you can get for Fortran? It certainly has helped me find bugs that gfortran, ifort and other have missed. You seem to be unaware of the fact that despite gfortran being free and quite good, there are at least five (that I can remember without googling) commercial Fortran compilers available today, e.g. from Nvidia/PGI and from Intel, that are quite popular in the scientific community.

Well, yes and no. Their product is by definition useless without a C compiler. Many (most?) people will use what's recommended, so many (most) people are using nagfor with gcc. A Core i7, on the other hand, works fine without a C compiler of any sort (as long as you only run precompiled binaries, which is true for (definitely this time) most people). But I see your point.

I'll give you an actual (but perhaps a little stretching-the-rules) example though:
NAG's Fortran compiler. It costs money. It's a Fortran-to-C compiler, and they recommend gcc if you don't have any other preferences. To me (as a compiler user) it's not a great difference if they make a Fortran-to-C compiler where I must provide the C compiler, or if they just shipped a C compiler and did the translation "in secret".

Well, why not? I found my right-hand little finger was cramping often. Coul be 'cause I use an "ergonomic" keyboard, which is quite good except the slightly bad position of the BackSpace key.

I see you've bound it to the Compose Key. That's utterly useless to me, since I very rarely write non-ascii characters (with the exception of æøå, and those I have dedicated keys for). When I want to write greek/fraktur/whatever, I write
$\alpha \to \infty \implies \sum_0^{\infty} n = -\frac{1}{12}$ etc. and then have the rendering engine (LaTeX) take care of it.

Quite. I find this page to be a fun insight into the world of stupid customers.

On my work keyboard, I've swapped CapsLock and Backspace. I occasionally get frustrated by it, but that's mainly when I ssh into that machine from a laptop at home. I think there's some mental association between "being on the work machine" and "having Backspace on the left hand".

Probably not. But their original papers are still too radioactive to touch without wearing protective equipment.

You could do the same already using scuba tanks made from carbon-fibre-titanium-nanotechnology (yeah I'm just adding buzzwords here). Those would be able to hold much more than 300 bars of pressure. You now why no-one's done that? Because during a normal plain-air dive with a 10 L 300 bar, you're not limited by the tank capacity. You're limited by how long you can stay submerged before you need to do decompression stops.

Yes, pure oxygen becomes toxic below 6 meters.

Very good point. Pure oxygen becomes toxic below 6 meters.

Also, looking at TFA and following the links, this looks like premium-class bullshit. No actual science, no pictures of the proposed device (just 3D renderings), this is just science-fiction.

That doesn't rule out the possibility that the sport only allows the helmets to be removed in cases where the risk of a serious collision is much lower (e.g. an ascent, like you mentioned, where the speed is much lower).