externals directory.  This is in particular useful because though
  most systems have bzip2/bunzip2, they don't always have libbz2,
  which we need for bsdiff/bspatch.

  derivation dependency graph.

  the `recurseForDerivations' attribute set to `true'.

  $ nix-build .../i686-linux.nix -A apacheHttpd -o apache

  components.

  packages (provided that they have a `meta.description' attribute).
  E.g.,

  $ ./src/nix-env/nix-env -qa --description gcc
  gcc-4.0.2   GNU Compiler Collection, 4.0.x (cross-compiler for sparc-linux)
  gcc-4.0.2   GNU Compiler Collection, 4.0.x (cross-compiler for mips-linux)
  gcc-4.0.2   GNU Compiler Collection, 4.0.x (cross-compiler for arm-linux)
  gcc-4.0.2   GNU Compiler Collection, 4.0.x

  re-allocations.

  to be queried, e.g., `nix-env -qa firefox'.  This does require the
  argument '*' to be passed if one wants information about all
  derivations, so the old `nix-env -qa' now is `nix-env -qa "*"'.

  ostringstreams.  Around 11% of execution time was spent here (now
  it's 0.5%).

  instantiation, e.g. "nix-env -i" and "nix-env -qas" (but not
  "nix-env -qa").  It turns out that many redundant calls to
  addToStore(path) were made, which reads and hashes the entire path.
  For instance, the bash bootstrap binary in Nixpkgs would be read and
  hashed many times.  As a result nix-env would spend around 92% of
  its time in the function sha256_block (according to callgrind).
  Some simple memoization fixes this.

  expressions that cause an assertion failure (like `assert system ==
  "i686-linux"').  This allows all-packages.nix in Nixpkgs to be used
  on all platforms, even if some Nix expressions don't work on all
  platforms.

  Not sure if this is a good idea; it's a bit hacky.  In particular,
  due to laziness some derivations might appear in `nix-env -qa' but
  disappear in `nix-env -qas' or `nix-env -i'.

  Commit 5000!

  Nix expression assertion failures.

  will be built or substituted.

  that arrays are a bash extension, so it didn't work on FreeBSD).
  Also fixes NIX-8 (readlink(1) dependency).

  base-32 hashes.

  parsed as 4.

  For a moment I worried that printHash32 was broken, and that would
  have been really, *really* bad ;-)