Commit graph

18825 commits

Author SHA1 Message Date
Max
1eb19dd804 libcmd: also include modify-installables.cc
Some checks failed
CI / vm_tests (push) Has been cancelled
CI / flake_regressions (push) Has been cancelled
CI / tests (macos-latest) (push) Has been cancelled
CI / docker_push_image (push) Has been cancelled
CI / tests (ubuntu-latest) (push) Has been cancelled
CI / Check Cachix and Docker secrets present for installer tests (push) Has been cancelled
CI / installer (push) Has been cancelled
CI / installer_test (macos-latest) (push) Has been cancelled
CI / installer_test (ubuntu-latest) (push) Has been cancelled
2024-11-10 21:40:32 +01:00
Max
b13b207102 Merge remote-tracking branch 'nixos/master' 2024-11-10 16:28:47 +01:00
Jörg Thalheim
aa9c0bc1ee
Merge pull request #11843 from xokdvium/dev/move-fixes
fix(treewide): clean up move semantics
2024-11-09 21:57:27 +01:00
Jörg Thalheim
cdcf9bd2fa
Merge pull request #11847 from xokdvium/dev/some-vector-reserves
refactor(treewide): reserve vector capacity when final size is known
2024-11-09 21:49:36 +01:00
Jörg Thalheim
6e55f2cf86
Merge pull request #11846 from xokdvium/dev/noexcept-more-ctors-where-possible
refactor(treewide): make some move ctors noexcept where appropriate
2024-11-09 21:48:26 +01:00
Sergei Zimmerman
0fe3b54ee1 refactor(treewide): reserve vector capacity when final size is known
In these trivial cases the final vector size (or lower bound on the size) is known,
so we can avoid some vector reallocations. This is not very important, but is just
good practice and general hygiene.
2024-11-09 22:40:49 +03:00
Sergei Zimmerman
96eeb6f4ff refactor(treewide): make some move ctors noexcept where appropriate
This is good practice to avoid pessimisations.
Left comments for the reasoning why ctors should be noexcept.
There are some tricky cases where we intentionally want throwing move ctors/assignments.
But those cases should really be reviewed, since some of those can be replaced
with more idiomatic copy/move-and-swap.
2024-11-09 12:09:33 +03:00
Sergei Zimmerman
0347bca15b fix(libstore/path-info): make ValidPathInfo move constructible/assignable 2024-11-08 20:48:00 +03:00
Sergei Zimmerman
149802b9f5 fix(libstore): make BasicDerivation move-constructible/assignable 2024-11-08 20:10:38 +03:00
Sergei Zimmerman
8dd787fbf6 fix(libutil): remove no-op move from const 2024-11-08 20:10:38 +03:00
Sergei Zimmerman
6c3f720e2c fix(treewide): move arguments where needed
Moving from arguments where it should be done.
2024-11-08 20:10:38 +03:00
Sergei Zimmerman
af63d67ba5 fix(libutils): make ref move assignable/constructible 2024-11-08 20:10:38 +03:00
Sergei Zimmerman
17b49134fa fix(treewide): fix incorrect usage of std::move
`auto &&` and `T &&` are forwarding references and can be
either lvalue or rvalue references. Moving from universal references
is incorrect and should not be done.

Moving from integral or floating-point values is pointless and just
worsens debug performance.
2024-11-08 20:10:38 +03:00
Jörg Thalheim
492c678162
Merge pull request #11837 from xokdvium/dev/fix-use-after-move-posix-source
fix(libutil/posix-source-accessor.cc): get rid of use-after-move bug
2024-11-08 14:56:50 +01:00
Sergei Zimmerman
3e0129ce3b fix(libutil/posix-source-accessor.cc): get rid of use-after-move bug
Naming class member variables the same as constructor arguments is a very
slippery slope because of how member variable names get resolved. Compiler
is not very helpful here and we need static analysis to forbid this kind of
stuff.

The following example illustrates the cause quite well:

```cpp

struct B {
    B(int) {}
};

struct A {
    A(int b): b([&](){
        return b;
        static_assert(std::is_same_v<decltype(b), int>);
    }()) {
       static_assert(std::is_same_v<decltype(b), int>);
    }
    void member() {
        static_assert(std::is_same_v<decltype(b), B>);
    }
    B b;
};

int main() {
    A(1).member();
}
```

From N4861 6.5.1 Unqualified name lookup:

> In all the cases listed in [basic.lookup.unqual], the scopes are searched
> for a declaration in the order listed in each of the respective categories;
> name lookup ends as soon as a declaration is found for the name.
> If no declaration is found, the program is ill-formed.

In the affected code there was a use-after-move for all accesses in the constructor
body, but this UB wasn't triggered.

These types of errors are trivial to catch via clang-tidy's [clang-analyzer-cplusplus.Move].
2024-11-08 16:07:45 +03:00
Eelco Dolstra
7630fe7c86
Merge pull request #11836 from DeterminateSystems/remove-shellInputs-i686-linux
Remove shellInputs.i686-linux
2024-11-08 13:52:13 +01:00
Eelco Dolstra
d5f5717172 Remove shellInputs.i686-linux
It fails to compile (https://hydra.nixos.org/build/277363696) and it's
unlikely anybody cares.
2024-11-08 13:14:43 +01:00
Jörg Thalheim
aeffdeffc8
Merge pull request #11826 from DeterminateSystems/revert-11804
Revert "Merge pull request #11804 from obsidiansystems/remove-old-make"
2024-11-08 12:04:55 +01:00
Eelco Dolstra
723fdeb4f1
Merge pull request #11832 from DeterminateSystems/fix-remote-build-test
Fix tests.remoteBuilds_local_nix_2_18
2024-11-07 22:56:28 +01:00
John Ericson
ff456f6713
Merge pull request #11833 from NixOS/fix-11830
Fix #11830
2024-11-07 16:08:55 -05:00
John Ericson
a6149eb89d Add eofOk parameter to the Windows readLine impl
Now the two implementations are back in sync.
2024-11-07 15:11:44 -05:00
John Ericson
372353722e Clean up standard stream logic
Now we have enough portability stuff
2024-11-07 15:11:41 -05:00
Eelco Dolstra
5471d2a454 Fix tests.remoteBuilds_local_nix_2_18
This was broken since a03bb4455c because
Nix 2.18 does not support broken $SHELL settings. So don't try a
broken $SHELL on old Nix versions. (It's a mystery though why
tests.remoteBuilds_local_nix_2_13 and tests.remoteBuilds_local_nix_2_3
didn't fail...)

https://hydra.nixos.org/build/277366807
2024-11-07 20:48:11 +01:00
Eelco Dolstra
0ed5af164f
Merge pull request #11831 from DeterminateSystems/fix-nix-copy-test
Fix 'nix copy' VM test
2024-11-07 20:43:01 +01:00
Eelco Dolstra
ea6dc8ebf1 Fix 'nix copy' VM test
This was broken because the root password wasn't getting set correctly.

https://hydra.nixos.org/build/277366631
2024-11-07 20:10:57 +01:00
Eelco Dolstra
fbceb864fa
Merge pull request #11829 from DeterminateSystems/revert-10980
Revert "Merge pull request #10980 from kjeremy/flake-show-description"
2024-11-07 20:10:24 +01:00
Eelco Dolstra
47bffe20e4 Revert "Merge pull request #10980 from kjeremy/flake-show-description"
This reverts commit 1c5ad159d6, reversing
changes made to 67de193277.

This reverts commit ce4e4a158e, reversing
changes made to 43e82c9446.
2024-11-07 18:12:04 +01:00
Eelco Dolstra
67d231c046 Revert "Merge pull request #11804 from obsidiansystems/remove-old-make"
This reverts commit 619eeb658a, reversing
changes made to 1af94bf471.
2024-11-07 13:46:37 +01:00
Jörg Thalheim
3ab8534b13
Merge pull request #11825 from hercules-ci/fix-arm-libatomic
Fix arm libatomic
2024-11-07 13:00:03 +01:00
Robert Hensing
f4b05cf8ec fix: Build nix-store on arm with libatomic 2024-11-07 11:40:02 +01:00
Robert Hensing
ffc1b30f50 refact: Extract build-utils-meson/libatomic 2024-11-07 11:40:02 +01:00
Jörg Thalheim
8f553f6eef
Merge pull request #11821 from hercules-ci/issue-11815
initRepoAtomically: Catch directory_not_empty
2024-11-07 07:22:23 +01:00
Robert Hensing
388271e8ec initRepoAtomically: Catch directory_not_empty 2024-11-06 23:52:48 +01:00
John Ericson
619eeb658a
Merge pull request #11804 from obsidiansystems/remove-old-make
Remove the old make-based build system
2024-11-06 16:41:25 -05:00
John Ericson
f7591bc6ed Make config.nix shims
As requested in https://github.com/NixOS/nix/pull/11792#discussion_r1827034428
2024-11-06 16:09:18 -05:00
John Ericson
e70c9bb06a Remove old build system 2024-11-06 16:09:18 -05:00
John Ericson
71c2e21f3b Temporarily skip coverage checks 2024-11-06 16:09:18 -05:00
John Ericson
ba34ea9fe7 Add missing deps to the dev shell 2024-11-06 16:09:18 -05:00
Robert Hensing
1af94bf471
Merge pull request #11816 from hercules-ci/fix-logger-ask-eof
`ProgressBar::ask`: accept EOF, as a no
2024-11-06 20:43:48 +01:00
Robert Hensing
deb1db8b7b
Merge pull request #11817 from DeterminateSystems/simplify-flake-regressions-sh
Move scripts/flake-regressions.sh into the flake-regressions repo
2024-11-06 20:40:14 +01:00
Robert Hensing
c6c8d2af65 refact: Remove unused arguments 2024-11-06 19:45:19 +01:00
Robert Hensing
3112e59734 ProgressBar::ask: Accept EOF as a no
This may occur when stderr is a tty but stdin is empty.
E.g.

    $ nix build </dev/null
    error: unexpected EOF reading a line

These stdio handles are how some non-interactive sandboxes behave,
including the Nix build sandbox and Hercules CI Effects.
2024-11-06 19:45:19 +01:00
Eelco Dolstra
c29808929f Move scripts/flake-regressions.sh into the flake-regressions repo
It already contained a script "eval-all.sh" that did almost the same thing.
2024-11-06 17:09:11 +01:00
Robert Hensing
f7b1e535a3 nix::readLine: Add eofOk parameter 2024-11-06 14:42:51 +01:00
Robert Hensing
4a785a0400
Merge pull request #11814 from xokdvium/dev/fix-libutil-tests-nix-api-util-leak
fix(libutil-tests/nix_api_util): get rid of unnecessary memory leaks
2024-11-06 12:36:00 +01:00
Sergei Zimmerman
2e7466a4e0 fix(libutil-tests/nix_api_util): get rid of unnecessary memory leaks 2024-11-06 10:55:39 +03:00
Jörg Thalheim
2ef5e222df
Merge pull request #11813 from xokdvium/dev/fix-use-after-free-libstore-tests
fix(libstore-tests): remove use-after-free bug for `StringSource`
2024-11-06 06:48:20 +01:00
John Ericson
bf19e5c34b
Merge pull request #11803 from obsidiansystems/factor-out-dev-shell
Dev shell improvements
2024-11-05 23:58:52 -05:00
John Ericson
c49a0ae8b8
Slightly tweak flake.nix
`lib.concatMapAttrs` instead of `lib.mapAttrs'` and `lib.nameValuePair`

Co-authored-by: Robert Hensing <roberth@users.noreply.github.com>
2024-11-05 23:22:43 -05:00
Sergei Zimmerman
5bc8957c73 fix(libstore-tests): remove use-after-free bug for StringSource
Unfortunately `StringSource` class is very easy was very easy to misuse
because the ctor took a plain `std::string_view` which has a bad habit
of being implicitly convertible from an rvalue `std::string`. This lead
to unintentional use-after-free bugs.

This patch makes `StringSource` much harder to misuse by disabling the ctor
from a `std::string &&` (but `const std::string &` is ok).

Fix affected tests from libstore-tests.
Reformat those tests with clangd's range formatting since the diff is tiny
and it seems appropriate.
2024-11-06 02:28:21 +03:00