I thought the article unfairly patronized Knuth too, but it makes an interesting point -- that if the kind of history Knuth wants to see is to continue, it will have to be done in Computer Science departments. It also provides examples of other disciplines, like mathematics, that have made space for historians internally.

I'd love to see this happen, but I don't know if it's politically viable within the academy.

I've never done it, but if you're managing that many FreeBSD servers you can use poudriere to build packages from ports with your preferred options and then deploy them as binaries. That should give you the best of both worlds.

I don't know why they still use sendmail. I wrote a sendmail configuration file once in my life: never again.

Also, you should check out the new package system if you haven't already. (It was called pkgng while in development; now it's just pkg, but pkgng is probably more googleable.) Ports still work fine, but there's been a lot of movement toward making binary packages the "cultural default" now that the tooling is better.

Indeed it has. Check out freebsd-update.

I wrote a toy Lisp interpreter in Objective-C, on Linux without any of the NS* libraries. It was great for that, because it gave polymorphism while at the same time letting you mess with object internals. My root class had a method that would overwrite the object's isa pointer, changing it into an object of another class, for use by the garbage collector; try that in most languages.

The basic layer is C, which is a small and relatively clean language, with a small and relatively clean Smalltalkish object model layered on top of it, with tools that let you bridge the gap when you need to do so. That's awesome. But Apple wants people to use it as an application language, and so they keep layering features and tying syntax to their runtime libraries to make up for the fact that it's a systems language at heart. Modern, Apple-influenced Objective-C is much less beautiful than the old language, though it may be more convenient for application development.

I don't follow the compilers very closely, so my recollection may be fuzzy, but I think this is what the grandparent was getting at:

In the past some people have proposed giving GCC a clear, serializable intermediate code so that one could use GCC front ends and back ends independently of the rest of GCC. The leadership viewed this as an end run around the GPL, because whatever produced or consumed this intermediate code would not need to link to the invoked portions of GCC. Tight coupling between the front ends and back ends was therefore treated as something to be encouraged for licensing reasons.

LLVM's design by contrast centers around its bytecode. Since it has a BSD-like license the question of whether you link to the compiler's code or fork/exec it is unimportant.

Since the emergence of LLVM, it's my understanding that the GCC leadership have softened their views -- the alternative to invoking GCC in a separate process is now to use LLVM, rather than to link to GCC and release one's source code; thus purely technical concerns can determine the design of GCC also.

If the car is truly autonomous and so competence of operation is not a concern, why should this be an automotive issue? If the parents are actually neglecting their children, deal accordingly.

It would make the most sense to require fewer qualifications as the technology becomes more proven; it could start requiring a driver's license with an endorsement and, as the cars become more capable and the kinks are worked out, go down to no license. But gradual deregulation tends to run counter to a bureaucracy's instincts and when the political process steps in it tends to do so suddenly, so I don't know if the idea would work in practice.

I agree that this is a silly concern for the average citizen; for policemen it's a real concern -- in that they often have to grapple with resisting people while their sidearms are in plain sight -- and it's telling that police departments are totally uninterested in smart guns.

This has never been the tradition in the English-speaking world. Some European countries have moved to this model since the nineteenth century.

His donation to Proposition 8 has been known literally for years. He was CTO before his promotion to CEO; he had a seat on the board of the Mozilla Foundation, as is natural for its co-founder. None of his gay subordinates or coworkers seems to have levied any accusations of unfairness against him in all that time. One of the Mozilla bigwigs commented that she was surprised to learn of the donation when it came out, because Eich's friendliness and evenhandedness toward gay employees defied her stereotype of a Proposition 8 supporter.

Eich had no trouble getting along with those who didn't share his views, but it seems that not everyone reciprocated.

Well yeah, that's because they added an essay portion that's scored separately. You'd expect the average score for the two-part test to be about 1,000 and the average score for the three-part test to be about 1,500.

They are offering the use of their money, which they could otherwise invest in some profitable enterprise; that is worth some interest. Without risk, one would expect it to be low interest -- like treasury bonds -- but certainly nonzero.

That's fair. I used to think I hated the blueness of common CFLs when what I actually hated was their terrible color rendering. I'm disappointed that there's not more of an emphasis on support for the full color spectrum in CFLs and LEDs.

Your tone is flamebait, but your question is valid. Firefox has a project called MemShrink whose focus has been on reducing memory usage. In the time they've been going they have found and fixed leaks in Firefox; come up with better ways to find leaks in add-ons, which were the biggest culprit; changed how Firefox handles memory used by add-ons to eliminate virtually all such leaks; and optimized Firefox's memory management in a bunch of non-buggy cases.

So yes, if memory usage is what drove you away from Firefox you should take another look.