With maximal laziness, you would expect that a function like this
fib = n:
if n == 0 then 0 else
if n == 1 then 1 else
builtins.add (fib (builtins.sub n 1)) (fib (builtins.sub n 2));
can be evaluated efficiently, because maximal laziness should
implictly memoize the recursive calls to "fib". However, non-strictness
interferes with this: the argument "n" is generally not in a form
that allows the memoization to work (e.g., it will be something like
(20 - 1 - 2 - 2) rather than 15). By the time that "n" is
evaluated (in "if n == 0 ..."), we're already deep in the evaluation
of the call.
(Strictness solves this:
builtins.add (strict fib (builtins.sub n 1)) (strict fib (builtins.sub n 2));
but that's not a very nice approach.)
With short-circuiting, the evaluator will check after evaluating a
term, whether that term is the argument of a function call that
we're currently evaluating. If so, it will check to see if the same
call but with the evaluated argument is in the normal form cache.
For instance, after evaluating (20 - 1 - 2 - 2) to 15, if we see
that "fib (20 - 1 - 2 - 2)" is currently being evaluated, we check
to see if "fib 15" is in the normal form cache. If so, we unwind
the stack (by throwing an exception) up to the evalExpr call
responsible for "fib (20 - 1 - 2 - 2)", which can then immediately
return the normal form for "fib 15". And indeed this makes "fib"
run in O(n) time.
The overhead for checking the active function calls (which isn't
very smart yet) seems to be modest, about 2% for "nix-env -qa
--drv-path --out-path" on Nixpkgs.
usage by finding identical files in the store and hard-linking them
to each other. It typically reduces the size of the store by
something like 25-35%. This is what the optimise-store.pl script
did, but the new command is faster and more correct (it's safe wrt
garbage collection and concurrent builds).
the given attribute path (just as -A does with other option)
(NIX-83). So you can now say
$ nix-env -qa -A nixpkgs_unstable.gnome \*
atk-1.12.4
esound-0.2.36
...
to see the packages in the "gnome" attribute in Nixpkgs.
To *print* the attribute path, you should now use "--attr-path" /
"-P" (running out of letters...).
Nix expressions in that directory are combined into an attribute set
{file1 = import file1; file2 = import file2; ...}, i.e. each Nix
expression is an attribute with the file name as the attribute
name. Also recurses into directories.
* nix-env: removed the "--import" (-I) option which set the
~/.nix-defexpr symlink.
* nix-channel: don't use "nix-env --import", instead symlink
~/.nix-defexpr/channels. So finally nix-channel --update doesn't
override any default Nix expressions but combines with them.
This means that you can have (say) a local Nixpkgs SVN tree and use
it as a default for nix-env:
$ ln -s .../path-to-nixpkgs-tree ~/.nix-defexpr/nixpkgs_svn
and be subscribed to channels (including Nixpkgs) at the same time.
(If there is any ambiguity, the -A flag can be used to
disambiguate, e.g. "nix-env -i -A nixpkgs_svn.pan".)
(/nix/var/nix/daemon-socket). This allows access to the Nix daemon
to be restricted by setting the mode/ownership on that directory as
desired, e.g.
$ chmod 770 /nix/var/nix/daemon-socket
$ chown root.wheel /nix/var/nix/daemon-socket
to allow only users in the wheel group to use Nix.
Setting the ownership on a socket is much trickier, since the socket
must be deleted and recreated every time the daemon is started
(which would require additional Nix configuration file directives to
specify the mode/ownership, and wouldn't support arbitrary ACLs),
some BSD variants appear to ignore permissions on sockets, and it's
not clear whether the umask is respected on every platform when
creating sockets.
fixed-output derivations or substitutions try to build the same
store path at the same time. Locking generally catches this, but
not between multiple goals in the same process. This happened
especially often (actually, only) in the build farm with fetchurl
downloads of the same file being executed on multiple machines and
then copied back to the main machine where they would clobber each
other (NIXBF-13).
Solution: if a goal notices that the output path is already locked,
then go to sleep until another goal finishes (hopefully the one
locking the path) and try again.
derivations that produce the same output path don't work properly
wrt locking. This happens a lot in the build farm when fetchurl
derivations downloading the same file on different platforms are
executed in parallel and then copied back to the main machine.
manifests directory. In that case, we don't do a nix-pull, so the
user gets pure source deployment.
The directory /nix/var/nix/gcroots/per-user/$USER should be
writable. (It's created automatically if
/nix/var/nix/gcroots/per-user is writable, e.g. if it has 1777
permission.)
need any info on substitutable paths, we just call the substituters
(such as download-using-manifests.pl) directly. This means that
it's no longer necessary for nix-pull to register substitutes or for
nix-channel to clear them, which makes those operations much faster
(NIX-95). Also, we don't have to worry about keeping nix-pull
manifests (in /nix/var/nix/manifests) and the database in sync with
each other.
The downside is that there is some overhead in calling an external
program to get the substitutes info. For instance, "nix-env -qas"
takes a bit longer.
Abolishing the substitutes table also makes the logic in
local-store.cc simpler, as we don't need to store info for invalid
paths. On the downside, you cannot do things like "nix-store -qR"
on a substitutable but invalid path (but nobody did that anyway).
* Never catch interrupts (the Interrupted exception).
NIX_DOWNLOAD_CACHE is set, then nix-prefetch-url will store the hash
and timestamp of downloaded files in the directory
$NIX_DOWNLOAD_CACHE. This allows it to figure out if the file is
still in the Nix store.
