The way we build python environments is subtly broken. A python
environment should be semantically identical to a vanilla Python
installation in, say, /usr/local. The current implementation, however,
differs in two important ways. The first is that it's impossible to use
python packages from the environment in python virtual environments. The
second is that the nix-generated environment appears to be a venv, but
it's not.
This commit changes the way python environments are built:
* When generating wrappers for python executables, we inherit argv[0]
from the wrapper. This causes python to initialize its configuration
in the environment with all the correct paths.
* We remove the sitecustomize.py file from the base python package.
This file was used tweak the python configuration after it was
incorrectly initialized. That's no longer necessary.
The end result is that python environments no longer appear to be venvs,
and behave more like a vanilla python installation. In addition it's
possible to create a venv using an environment and use packages from
both the environment and the venv.
We also renamed `filter` (as a name of a parameter) to `predicate` following the naming suggestion in code review. It's better!
Since it's not part of an attrset, the name can change with no impact to semantics, since it can't be observed with `builtins.functionArgs`.
```
$ nix-repl
Nix 2.21.0
Type :? for help.
nix-repl> f = x: y: z: (x + y + z)
nix-repl> builtins.functionArgs f
{ }
nix-repl> :doc builtins.functionArgs
Synopsis: builtins.functionArgs f
Return a set containing the names of the formal arguments expected by the function f. The value of each attribute is a Boolean denoting whether the corresponding argument has a default value. For instance, functionArgs ({ x, y ?
123}: ...) = { x = false; y = true; }.
"Formal argument" here refers to the attributes pattern-matched by the function. Plain lambdas are not included, e.g. functionArgs (x: ...) = { }.
```
The nixpkgs-unstable channel's programs.sqlite was used to identify
packages producing exactly one binary, and these automatically added
to their package definitions wherever possible.
The `-path` test of `find` does string comparison, not path comparison.
Hence, the format of `everythingFile` needed to be very specific. Now,
it can be denormalized (e.g. it can contain `/./`) and an error is
emitted if the everything file or its interface file can't be removed.
Upstream now provides a library file for the builtin library and ensured
that the existing interface files will be used regardless of whether
--local-interfaces is in effect. Hence, Agda will not try to write to
the Nix store anymore except if the build flags are changed.
Replace writeReferencesToFile with writeClosure.
Make writeClosure accept a list of paths instead of a path.
Re-implement with JSON-based exportReferencesGraph interface provided by
__structuredAttrs = true.
Reword the documentation.
Co-authored-by: Valentin Gagarin <valentin.gagarin@tweag.io>
Co-authored-by: Robert Hensing <roberth@users.noreply.github.com>
Co-authored-by: Someone Serge <sergei.kozlukov@aalto.fi>
In NixOS/nixpkgs#290081 it came to attention that autoPatchelfHook is
one of if not the only hook in Nixpkgs that is a multiline string
expression. Almost all hooks are functions, which guard with something
like `if [ -z "${dontDoTheThing-}" ]; then ...` in the function, or
single-line strings which include that guard inline and then call the
real function, e.g. `if [ -z "${dontDoTheThing-} ]; then doTheThing; fi`.
This commit moves autoPatchelfHook to the former, which seems to be the
most common style now.
- Use testers.runNixOSTest instead of testers.nixosTest as nixosTest
has become obsolete.
- Prepare for cross-platform testing.
- Use the testScriptBin passthru'd by the references test package
inside the guest pkgs.
swiftc uses cc-wrapper which sets the -march flag on some systems which
breaks the build. This change adds a flag, disableMarch, to cc-wrapper
which disables using the -march flag.
https://github.com/NixOS/nixpkgs/issues/295322
This assumes that downstream users of `buildNpmPackage` would rather our
own built `node_modules` be copied to the output rather than only the
`bundleDependencies` specified in the `package.json` file.
Having the latter behavior seems unexpected and unintuitive, and would
not work as installing from an `npm pack` is intended to (since doing
that would not do a `rebuild` step on those dependencies and it would
skip reifying a full dependency tree).
- merge libcxxabi into libcxx for LLVM 12, 13, 14, 15, 16, 17, and git.
- remove the link time workaround `-lc++ -lc++abi` from 58 packages as it is no longer required.
- fixes https://github.com/NixOS/nixpkgs/issues/166205
- provides alternative fixes for. https://github.com/NixOS/nixpkgs/issues/269548https://github.com/NixOS/nix/issues/9640
- pkgsCross.x86_64-freebsd builds work again
This change can be represented in 3 stages
1. merge libcxxabi into libcxx -- files: pkgs/development/compilers/llvm/[12, git]/{libcxx, libcxxabi}
2. update stdenv to account for merge -- files: stdenv.{adapters, cc.wrapper, darwin}
3. remove all references to libcxxabi outside of llvm (about 58 packages modified)
### merging libcxxabi into libcxx
- take the union of the libcxxabi and libcxx cmake flags
- eliminate the libcxx-headers-only package - it was only needed to break libcxx <-> libcxxabi circular dependency
- libcxx.cxxabi is removed. external cxxabi (freebsd) will symlink headers / libs into libcxx.
- darwin will re-export the libcxxabi symbols into libcxx so linking `-lc++` is sufficient.
- linux/freebsd `libc++.so` is a linker script `LINK(libc++.so.1, -lc++abi)` making `-lc++` sufficient.
- libcxx/default.nix [12, 17] are identical except for patches and `LIBCXX_ADDITIONAL_LIBRARIES` (only used in 16+)
- git/libcxx/defaul.nix does not link with -nostdlib when useLLVM is true so flag is removed. this is not much different than before as libcxxabi used -nostdlib where libcxx did not, so libc was linked in anyway.
### stdenv changes
- darwin bootstrap, remove references to libcxxabi and cxxabi
- cc-wrapper: remove c++ link workaround when libcxx.cxxabi doesn't exist (still exists for LLVM pre 12)
- adapter: update overrideLibcxx to account for a pkgs.stdenv that only has libcxx
### 58 package updates
- remove `NIX_LDFLAGS = "-l${stdenv.cc.libcxx.cxxabi.libName}` as no longer needed
- swift, nodejs_v8 remove libcxxabi references in the clang override
https://github.com/NixOS/nixpkgs/pull/292043
Add a makeWrapperArgs argument to all script writers under pkgs.writers.
This can be used to set, prefix, or suffix the PATH or other environment variables which improves the ability to generate scripts with reproducible behavior.
Some of the writers (writeBash, writeDash, writeFish, writeNu) previously did not support passing an argument set, for example
```
writeBash "example" "echo hello"
```
In order to add the new capability to these writers as well, their call signature is now overloaded in order to allow the following:
(The old call style from the example above remains intact)
```
writeBash "example"
{ makeWrapperArgs = [ "--prefix" "PATH" ":" "${pkgs.hello}/bin" ]; }
''
hello
''
```
Done as well:
- add tests
- add more docs
- fix some misleading docs
- extend existing docs with more examples
`snapTools.makeSnap` has produced broken snaps since at least Oct 2020,
as indicated by the following issue: https://github.com/NixOS/nixpkgs/issues/100618
No person has shown interest in maintaining it, and given that there is
no fix available, it's assumed that all attempts made to fix that
function have not succeeded.
Given that `snapTools` only contained `makeSnap`, it was removed
completely.
This is an alternative to `fetchNpmDeps` that is notably different in that it uses metadata from `package.json` & `package-lock.json` instead of specifying a fixed-output hash.
Notable features:
- IFD free.
- Only fetches a node dependency once. No massive FODs.
- Support for URL, Git and path dependencies.
- Uses most of the existing `npmHooks`
`importNpmLock` can be used _only_ in the cases where we need to check in a `package-lock.json` in the tree.
Currently this means that we have 13 packages that would be candidates to use this function, though I expect most usage to be in private repositories.
This is upstreaming the builder portion of https://github.com/adisbladis/buildNodeModules into nixpkgs (different naming but the code is the same).
I will archive this repository and consider nixpkgs the new upstream once it's been merged.
For more explanations and rationale see https://discourse.nixos.org/t/buildnodemodules-the-dumbest-node-to-nix-packaging-tool-yet/35733
Example usage:
``` nix
stdenv.mkDerivation {
pname = "my-nodejs-app";
version = "0.1.0";
src = ./.;
nativeBuildInputs = [
importNpmLock.hooks.npmConfigHook
nodejs
nodejs.passthru.python # for node-gyp
npmHooks.npmBuildHook
npmHooks.npmInstallHook
];
npmDeps = buildNodeModules.fetchNodeModules {
npmRoot = ./.;
};
}
```
We need to set -crt-static on musl for rustdoc as well, so let's unify
the wrappers. Ideally, rather than wrapping rustdoc, we'd have
rustdoc use the wrapped rustc, but that's currently only possible with
an unstable option (--test-builder).
The options set by the wrapper, -C target-feature and --sysroot, are
supported by both rustdoc and rustc, but other flags maybe not be
supported by both, so I've introduced different environment
variables (the existing NIX_RUSTFLAGS and a new NIX_RUSTDOCFLAGS) to
allow those to be set independently.
This fixes cargo-auditable in pkgsMusl., which broke because its
doctests stopped working when -crt-static was moved to the wrapper.
Fixes: 79156bf13a ("rustc: move crt-static default override to wrapper (#291829)")
Closes#193336Closes#261694
Related to #108984
The goal here was to get the following flake to build and run on
`aarch64-darwin`:
```nix
{ inputs.nixpkgs.url = <this branch>;
outputs = { nixpkgs, ... }: {
checks.aarch64-darwin.default =
nixpkgs.legacyPackages.aarch64-darwin.nixosTest {
name = "test";
nodes.machine = { };
testScript = "";
};
};
}
```
… and after this change it does. There's no longer a need for the
user to set `nodes.*.nixpkgs.pkgs` or
`nodes.*.virtualisation.host.pkgs` as the correct values are inferred
from the host system.
If -target is explicitly passed to clang, we shouldn't pass our -march
value for the default target, because it probably won't exist for the
target being used. Up until now, clang has been lenient with this,
but it's a hard error with clang 17, so since gcc.arch is always set
on aarch64, fixing this is a hard requirement for upgrading our
default clang to 17.
Before (with clang 17 on aarch64-linux):
$ clang -target bpf -c -o /dev/null test.bpf.c
clang: warning: ignoring '-fstack-protector-strong' option as it is not currently supported for target 'bpf' [-Woption-ignored]
clang: error: unsupported option '-march=' for target 'bpf'
clang: warning: argument unused during compilation: '--gcc-toolchain=/nix/store/cngak08nb1yk44gnjipv5rg1ahh1xkax-gcc-13.2.0' [-Wunused-command-line-argument]
After:
$ clang -target bpf -c -o /dev/null test.bpf.c
clang: warning: ignoring '-fstack-protector-strong' option as it is not currently supported for target 'bpf' [-Woption-ignored]
clang: warning: argument unused during compilation: '--gcc-toolchain=/nix/store/cngak08nb1yk44gnjipv5rg1ahh1xkax-gcc-13.2.0' [-Wunused-command-line-argument]
Add a makeWrapperArgs argument to all script writers under pkgs.writers.
This can be used to set, prefix, or suffix the PATH or other environment variables which improves the ability to generate scripts with reproducible behavior.
Some of the writers (writeBash, writeDash, writeFish, writeNu) previously did not support passing an argument set, for example
```
writeBash "example" "echo hello"
```
In order to add the new capability to these writers as well, their call signature is now overloaded in order to allow the following:
(The old call style from the example above remains intact)
```
writeBash "example"
{ makeWrapperArgs = [ "--prefix" "PATH" ":" "${pkgs.hello}/bin" ]; }
''
hello
''
```
Done as well:
- add tests
- add more docs
- fix some misleading docs
- extend existing docs with more examples
Previously, when cross compiling from non-musl to musl, the crt-static
default override wouldn't be applied, because the compiler wouldn't
have been built with it due to fastCross. Moving it to the wrapper
fixes this without having to introduce extra compiler rebuilds. And
because the wrapper is applied even to the bootstrap rustc, we no
longer need special handling of crt-static in the Cargo expression.
Unlike --sysroot, rustc allows -C target-feature= to be passed
multiple times, with later instances taking precedence over earlier
ones. This means that it's very easy to set the default in the
wrapper, just by our overridden default before any other arguments.
This fixes pkgsCross.aarch64-multiplatform-musl.mesa from x86_64-linux.
The change is insignificant when the owner is root. However, when it
is not root, this change is needed to allow using Nix (as an
unprivileged user) inside the container.
Ada depencencies musst be build with the same gnat version as the
project. Use a namespace as preperation to build with different gnat
versions.
gprbuild and gnatprove are still globaly visable.
this equates to -ftrivial-auto-var-init=pattern
clang has removed support for -ftrivial-auto-var-init=zero and
are unlikely to re-add it, so use -ftrivial-auto-var-init=pattern
on both compilers if only to make behaviour more consistent
between the two.
add to pkgsExtraHardening's defaultHardeningFlags.
The loaded database contains timestamps of when the nix paths were
registered. Depending on the host store, these can differ between runs.
Resetting them to a well known values ensures that the produced image is
reproducible.
This commit adds support for swapping out the compression algorithm
used in all major docker-tools commands that generate images. The
default algorithm remains unchanged (gzip).
This fetcher is to be used with PyPi mirrors exposing the "legacy" API, such as devpi.
A variant of this fetcher has been used in poetry2nix for years and
has served us well there to support private PyPi mirrors and Devpi.
Example usage:
``` nix
fetchPypiLegacy {
file = "urllib3-1.26.2.tar.gz";
hash = "sha256:19188f96923873c92ccb987120ec4acaa12f0461fa9ce5d3d0772bc965a39e08";
pname = "urllib3";
url = "https://pypi.org/simple";
}
```
cc @lewo who wrote the this originally
cc contributors @rskew @gmacon @jperras @Smaug123
Without the change tests like `xorg.imake.tests.pkg-config` fail as:
$ nix build --no-link -f. -L xorg.imake.tests.pkg-config
error: builder for '/nix/store/i3zb1ykjzm0622497cn4dvifk36sx00r-check-pkg-config-.drv' failed to produce output path for output 'out' at '/nix/store/i3zb1ykjzm0622497cn4dvifk36sx00r-check-pkg-config-.drv.chroot/nix/store/63y92rvkp7gzzp0hlcjyj92srzjwndrq-check-pkg-config-'
error: 1 dependencies of derivation '/nix/store/8qycf49a4h9jj2662d9cf4d56aq6djjj-check-meta-pkg-config-modules-for-imake-1.0.9.drv' failed to build
Let's always produce empty output for zero-modules tests.
The previous version didn't work in the case of relative symlinks in
subdirectories. If "foo/bar" was a link to "baz", it would check for
a link to "baz" in the root, rather than under "foo".
We don't need to dereference the symlink ourselves for [ anyway, as it
dereferences its arguments itself, so all we need to do to fix this is
to pass it the link.
Fixes: 14f83d5c6f ("compressFirmwareXz: fix links to directories")
`multiPaths` is defined via invalid operation on `null` value for
non-multilib environments. Noticed on `notesnook` evan failure as:
nix-repl> notesnook.fhsenv.multiPaths
253| passthru = {
254| inherit args baseTargetPaths targetPaths baseMultiPaths multiPaths ldconfig;
| ^
255| };
error: attempt to call something which is not a function but null
The change makes `multiPaths` a private variable.
The unwrapped version doesn't know where to look for libraries, so
this is required to e.g. build anything that uses openssl-sys with
OPENSSL_NO_VENDOR. A randomly chosen example package that's fixed by
this change is pkgsStatic.gitoxide.
We're already using pkgsBuildBuild, and we'll soon be using
pkgsBuildTarget, so for consistency, change buildPackages and
targetPackages to their corresponding two-component names.
Pass the whole environment to the native-image build process by
generating a -E option for every environment variable.
This has the same effect as setting
NATIVE_IMAGE_DEPRECATED_BUILDER_SANITATION=true
but is compatible with packages providing -E options themselves
Packages that require access to environment variables in the build
should specify these using `-E` arguments in `nativeImageBuildArgs` or
using a `native-image.properties` as described here:
https://www.graalvm.org/22.1/reference-manual/native-image/BuildConfiguration/#embedding-a-configuration-file
Specifying NATIVE_IMAGE_DEPRECATED_BUILDER_SANITATION=true breaks with
packages that list their environment variables explicitly in `native-image.properties`.
this allows a compiler derivation to provide a
hardeningUnsupportedFlagsByTargetPlatform passthru attr
that will be called with the targetPlatform to determine
the unsupported hardening flags for that platform.
we can do this because even though a clang compiler is
multi-target by nature, cc-wrapper effectively fixes the
target platform at wrapping time. otherwise we'd have to
sniff the intended target at runtime, which wouldn't
be fun at all.
the advantage of using a new attribute instead of
allowing hardeningUnsupportedFlags to optionally be a
function is that hardeningUnsupportedFlags retains its
simple overriding pattern for simple cases (i.e.
`(prev.hardeningUnsupportedFlags or []) ++ [ "foo" ]`
) which will continue to work as long as the bottom-most
function of hardeningUnsupportedFlagsByTargetPlatform
falls back to hardeningUnsupportedFlags.
PR #275947, which was self-merged without approvals, inserted
functionality specific to a propriteary closed-source compiler
(CUDA) into cc-wrapper.
This commit relocates this CUDA-specific functionality into the
appropritate place: `cuda-modules`.
It is unclear to me exactly what this function is supposed to be
doing; much of it (like the `.kind` attributes) do not appear to be
used *anywhere* in nixpkgs. Making sure we don't insert unexplained
deadcode like this is one of the important functions of the review
process.
This commit deletes speculative comments which were self-merged with
no approvals in PR #275947.
If you think that "The above 'fix' may be incorrect" the correct
response is to submit a PR which removes the 'fix' and get it reviewed.
Likewise, if you think that "For clang it's not necessary" you
should submit a PR which wraps it in `if !isClang`.
`cc-wrapper` is full of too much junk as it is, let's not make
things worse.
The commit prior to this one, "gcc: fix c++ headers when same
triplet cross compiling" causes gcc's c++ headers to be in the same
outpath subdirectory regardless of whether the gcc build is a
host==target or host!=target compiler.
As a result of that change, the hack in cc-wrapper which adapted to
the different paths is no longer needed. And, in fact, it must be
removed, since if it is left in place builds such as
pkgsCross.aarch64-multiplatform.firefox will fail as shown below.
```
firefox-unwrapped> 0:02.01(B checking the host C compiler works... yes(B(B
firefox-unwrapped> 0:02.01(B checking for the host C++ compiler... /nix/store/1asqji9djmdlapzs70q7jw2j308ry7cn-clang-wrapper-16.0.6/bin/c++(B(B
firefox-unwrapped> 0:02.14(B checking whether the host C++ compiler can be used... yes(B(B
firefox-unwrapped> 0:02.14(B checking the host C++ compiler version... 16.0.6(B(B
firefox-unwrapped> 0:02.21(B checking the host C++ compiler works... yes(B(B
firefox-unwrapped> 0:02.40(B checking for target linker... lld(B(B
firefox-unwrapped> 0:02.51(B checking for host linker... lld(B(B
firefox-unwrapped> 0:02.60(B checking for 64-bit OS... yes(B(B
firefox-unwrapped> 0:02.67(B checking for new enough STL headers from libstdc++...(B(B
firefox-unwrapped> 0:02.67(B DEBUG: <truncated - see config.log for full output>(B(B
firefox-unwrapped> 0:02.67(B DEBUG: | #if defined(__GLIBCXX__) && !defined(_GLIBCXX_RELEASE)(B(B
firefox-unwrapped> 0:02.67(B DEBUG: | # error libstdc++ not new enough(B(B
firefox-unwrapped> 0:02.67(B DEBUG: | #endif(B(B
firefox-unwrapped> 0:02.67(B DEBUG: | #if defined(_GLIBCXX_RELEASE)(B(B
firefox-unwrapped> 0:02.67(B DEBUG: | # if _GLIBCXX_RELEASE < 8(B(B
firefox-unwrapped> 0:02.67(B DEBUG: | # error libstdc++ not new enough(B(B
firefox-unwrapped> 0:02.67(B DEBUG: | # else(B(B
firefox-unwrapped> 0:02.67(B DEBUG: | (void) 0(B(B
firefox-unwrapped> 0:02.67(B DEBUG: | # endif(B(B
firefox-unwrapped> 0:02.67(B DEBUG: | #endif(B(B
firefox-unwrapped> 0:02.67(B DEBUG: | ;(B(B
firefox-unwrapped> 0:02.67(B DEBUG: | return 0;(B(B
firefox-unwrapped> 0:02.67(B DEBUG: | }(B(B
firefox-unwrapped> 0:02.67(B DEBUG: Executing: `/nix/store/7v4bi4q334yircaznwm353h1l5i7k98f-aarch64-unknown-linux-gnu-clang-wrapper-16.0.6/bin/aarch64-unknown-linux-gnu-clang++ /build/conftest.qvbo58dv.cpp -c`(B(B
firefox-unwrapped> 0:02.67(B DEBUG: The command returned non-zero exit status 1.(B(B
firefox-unwrapped> 0:02.67(B DEBUG: Its error output was:(B(B
firefox-unwrapped> 0:02.67(B DEBUG: | /build/conftest.qvbo58dv.cpp:1:10: fatal error: 'cstddef' file not found(B(B
firefox-unwrapped> 0:02.67(B DEBUG: | #include <cstddef>(B(B
firefox-unwrapped> 0:02.67(B DEBUG: | ^~~~~~~~~(B(B
firefox-unwrapped> 0:02.67(B DEBUG: | 1 error generated.(B(B
firefox-unwrapped> 0:02.67(B ERROR: The libstdc++ in use is not new enough. Please run ./mach bootstrap to update your compiler, or update your system libstdc++ installation.(B(B
firefox-unwrapped> *** Fix above errors and then restart with "./mach build"
error: build of '/nix/store/5in7xkji5hzqkl14ygwq3vxnni54lykk-firefox-unwrapped-aarch64-unknown-linux-gnu-119.0.1.drv' on 'ssh://root@192.168.22.103' failed: builder for '/nix/store/5in7xkji5hzqkl14ygwq3vxnni54lykk-firefox-unwrapped-aarch64-unknown-linux-gnu-119.0.1.drv' failed with exit code 1
error: builder for '/nix/store/5in7xkji5hzqkl14ygwq3vxnni54lykk-firefox-unwrapped-aarch64-unknown-linux-gnu-119.0.1.drv' failed with exit code 1;
```
* doc: improve documentation for trivial text writing functions
Co-authored-by: Brian Merchant <bzm3r@proton.me>
Co-authored-by: Robert Hensing <robert@roberthensing.nl>
Co-authored-by: Alexander Groleau <alex@proof.construction>
Co-authored-by: Valentin Gagarin <valentin.gagarin@tweag.io>
