Progress on #5638
There are still a global fetcher and eval settings, but they are pushed
down into `libnixcmd`, which is a lot less bad a place for this sort of
thing.
Continuing process pioneered in
52bfccf8d8.
It is unclear to me why this worked when not in a VM test, but the
explanation would be in the part of nix-shell we're getting rid of
with the devShell attribute.
When --unpack was used the nix would add the current directory to the
nix store instead of the content of unpacked.
The reason for this is that std::distance already consumes the iterator.
To fix this we re-instantiate the directory iterator in case the
directory only contains a single entry.
- use the iterator in `CanonPath` to count `level`
- use the `CanonPath::basename` method
- use `CanonPath::root` instead of `CanonPath{""}`
- remove `Path` and `PathView`, use `std::filesystem::path` directly
Inspired by
010ff57ebb
From the original PR:
> We do not have any of these warnings appearing at the moment, but
> it seems like a good idea to enable [[nodiscard]] checking anyway.
> Once we start introducing more functions with must-use conditions we will
> need such checking, and the rust stdlib has proven them very useful.
The code that counts the number of elided attrs incorrectly used the
per-printer "global" attribute counter instead of a counter that
was relevant only to the current attribute set.
This bug flew under the radar because often the attribute sets aren't
nested, not big enough, or we wouldn't pay attention to the numbers.
I've noticed the issue because the difference underflowed.
Although this behavior is tested by the functional test
lang/eval-fail-bad-string-interpolation-4.nix, the underflow slipped
through review. A simpler reproducer would be as follows, but I
haven't added it to the test suite to keep it simple and marginally
faster.
```
$ nix run nix/2.23.1 -- eval --expr '"" + (let v = { a = { a = 1; b = 2; c = 1; d = 1; e = 1; f = 1; g = 1; h = 1; }; b = { a = 1; b = 1; c = 1; }; }; in builtins.deepSeq v v)'
error:
… while evaluating a path segment
at «string»:1:6:
1| "" + (let v = { a = { a = 1; b = 2; c = 1; d = 1; e = 1; f = 1; g = 1; h = 1; }; b = { a = 1; b = 1; c = 1; }; }; in builtins.deepSeq v v)
| ^
error: cannot coerce a set to a string: { a = { a = 1; b = 2; c = 1; d = 1; e = 1; f = 1; g = 1; h = 1; }; b = { a = 1; «4294967289 attributes elided» }; }
```
The old `std::variant` is bad because we aren't adding a new case to
`FileIngestionMethod` so much as we are defining a separate concept ---
store object content addressing rather than file system object content
addressing. As such, it is more correct to just create a fresh
enumeration.
Co-authored-by: Robert Hensing <roberth@users.noreply.github.com>
This tests the parser and JSON format using the DRV files from the tests
added in the previous commit.
Co-Authored-By: John Ericson <John.Ericson@Obsidian.Systems>
This tests the Nix language side of things.
We are purposely skipping most of `common.sh` because it is overkill for
this test: we don't want to have an "overfit" test environment.
Co-Authored-By: John Ericson <John.Ericson@Obsidian.Systems>
Instead of running the builds under
`$TMPDIR/{unique-build-directory-owned-by-the-build-user}`, run them
under `$TMPDIR/{unique-build-directory-owned-by-the-daemon}/{subdir-owned-by-the-build-user}`
where the build directory is only readable and traversable by the daemon user.
This achieves two things:
1. It prevents builders from making their build directory world-readable
(or even writeable), which would allow the outside world to interact
with them.
2. It prevents external processes running as the build user (either
because that somehow leaked, maybe as a consequence of 1., or because
`build-users` isn't in use) from gaining access to the build
directory.
In most real world cases, the Link header is set on the redirect, not on
the final file. This regressed in Lix earlier and while new unit tests
were added to cover it, this integration test should probably have also
caught it.
Source: a3256a9375
... so that we may perhaps later extend the interface.
Note that Nixpkgs' lib.warn already requires a string coercible
argument, so this is reasonable. Also note that string coercible
values aren't all strings, but in practice, for warn, they are.
Fixes assertion failure if outputsToInstall is empty by defaulting to the "out"
output. That is, behavior between the following commands should be consistent:
$ nix build --no-link --json .#nothing-to-install-no-out
error: derivation '/nix/store/eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee-nothing-to-install-no-out.drv' does not have wanted outputs 'out'
$ nix build --no-link --file default.nix --json nothing-to-install-no-out
error: derivation '/nix/store/eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee-nothing-to-install-no-out.drv' does not have wanted outputs 'out'
Real-world example of this issue:
$ nix build --json .#.legacyPackages.aarch64-linux.texlive.pkgs.iwona
error: derivation '/nix/store/dj0h6b0pnlnan5nidnhqa0bmzq4rv6sx-iwona-0.995b.drv' does not have wanted outputs 'out'
$ git rev-parse HEAD
eee33247cf6941daea8398c976bd2dda7962b125
$ nix build --json --file . texlive.pkgs.iwona
nix: src/libstore/outputs-spec.hh:46: nix::OutputsSpec::Names::Names(std::set<std::__cxx11::basic_string<char> >&&): Assertion `!empty()' failed.
Aborted (core dumped)
This increases test coverage, and gets the worker protocol ready to be
used by Hydra.
Why don't we just try to use the store interface in Hydra? Well, the
problem is that the store interface works on connection pools, with each
opreation getting potentially a different connection, but the way temp
roots work requires that we keep one logical "transaction" (temp root
session) using the same connection.
The longer-term solution probably is making connections themselves
implement the store interface, but that is something that builds on
this, so I feel OK that this is not churn in the wrong direction.
Fixes#9584
Do this instead of an unchecked cast
I redid this to use the serialisation framework (including a unit test),
but I am keeping the reference to credit Jade for spotting the issue.
Change-Id: Icf6af7935e8f139bef36b40ad475e973aa48855c
(adapted from commit 2a7a824d83dc5fb33326b8b89625685f283a743b)
Co-Authored-By: Jade Lovelace <lix@jade.fyi>
This way we can commit the same amount of stack size (64 MB) without a conditional.
Includes nix, libnixexpr-tests, libnixfetchers-tests, libnixstore-tests, libnixutil-tests.
https://github.com/NixOS/nix/pull/10555 added a check requiring
that output parameters always have an uninitialized Value as argument.
Unfortunately the output parameter of the primop callback received
a thunk instead.
See the comment for implementation considerations.
This was accidentally removed in
e989c83b44. I restored it and also did a
few other cleanups:
- Make a static method for namespacing purposes
- Put the test files in the data dir with the other test data
- Avoid mutating globals in the machine config tests
This will be used by Hydra.
In particular `local://<path>` and `unix://` (without any path) now
work, and mean the same things as `local` and `daemon`, respectively. We
thus now have the opportunity to desguar `local` and `daemon` early.
This will allow me to make a change to
https://github.com/NixOS/nix/pull/9839 requested during review to
desugar those earlier.
Co-authored-by: Théophane Hufschmitt <7226587+thufschmitt@users.noreply.github.com>
1. Hydra currently queries for multiple path infos at once, so let us
make a connection item for that.
2. The minimum of the two versions should always be used, see #9584.
(The issue remains open because the daemon protocol needs to be
likewise updated.)
The JSON format no longer uses the legacy ATerm `r:` prefixing nonsese,
but separate fields.
Progress on #9866
Co-authored-by: Robert Hensing <roberth@users.noreply.github.com>
Basically I'd expect the same behavior as with `nix-build`, i.e.
with `--keep-going` the hash-mismatch error of each failing
fixed-output derivation is shown.
The approach is derived from `Store::buildPaths` (`entry-point.cc`):
instead of throwing the first build-result, check if there are any build
errors and if so, display all of them and throw after that.
Unfortunately, the BuildResult struct doesn't have an `ErrorInfo`
(there's a FIXME for that at least), so I have to construct my own here.
This is a rather cheap bugfix and I decided against touching too many
parts of libstore for that (also I don't know if that's in line with the
ongoing refactoring work).
Closes https://git.lix.systems/lix-project/lix/issues/302
Change-Id: I378ab984fa271e6808c6897c45e0f070eb4c6fac
Signed-off-by: Jörg Thalheim <joerg@thalheim.io>
previously the test directory could have been left untouched before executing
a test when `init.sh` was not run - and sometimes it isn't
supposed to be run - which made the test suite highly stateful and thus
behaving surprisingly on multiple runs.
pararameterisation is not actually needed the way things are currently
set up, and it confused me when trying to understand what the code does.
all but one test sources vars-and-functions.sh, which nominally only
defines variables, but in practice is always coupled with the actual
initialisation. while the cleaner way of making this more legible would
be to source variables and initialisation separately, this would produce
a huge diff.
the change requires a few small fixes to keep the tests working:
- only create test home directory during initialisation
that vars-and-functions.sh wrote to the file system seems not write
- fix creation of the test directory
due to statefulness, the test home directory was implicitly creating
the test root, too. decoupling that made it apparent that this was
probably not intentional, and certainly confusing.
- only source vars-and-functions.sh if init.sh is not needed
there is one test case that only needs a helper function but no
initialisation side effects
- remove some unnecessary cleanups and split parts of re-used test code
there were confusing bits in how initialisation code was repurposed,
which break if trying to refactor the outer layers naively...
