Commit graph

10232 commits

Author SHA1 Message Date
Eelco Dolstra
a878c448d8
Merge pull request #4551 from danieldk/system-features-compute-level
Add x86_64 compute levels as system features
2021-02-23 14:30:10 +01:00
regnat
ba1a256d08 Make DerivationGoal::drv a full Derivation
This field used to be a `BasicDerivation`, but this `BasicDerivation`
was downcasted to a `Derivation` when needed (implicitely or not), so we
might as well make it a full `Derivation` and upcast it when needed.

This also allows getting rid of a weird duplication in the way we
compute the static output hashes for the derivation. We had to
do it differently and in a different place depending on whether the
derivation was a full derivation or just a basic drv, but we can now do
it unconditionally on the full derivation.

Fix #4559
2021-02-23 14:15:45 +01:00
regnat
c32168c9bc Test the remote building of ca derivations 2021-02-23 08:04:03 +01:00
regnat
527da73690 Properly bypass the registering step when all outputs are present
There was already some logic for that, but it didn't handle the case of
content-addressed outputs, so extend it a bit for that
2021-02-23 08:04:03 +01:00
regnat
69666b951e build-remote: Always register the missing outputs
It's possible that all the paths are already there, but just not
associated to the current drv output
2021-02-23 08:04:03 +01:00
regnat
8c385d16ee Also send ca outputs to the build hook
Otherwise they don't get registered, triggering an assertion failure
at some point later
2021-02-23 08:04:03 +01:00
regnat
27b5747ca7 RemoteStore: Send back the new realisations
To allow it to build ca derivations remotely
2021-02-23 08:04:03 +01:00
regnat
a2b69660a9 LegacySSHStore: Send back the new realisations
To allow it to build ca derivations remotely
2021-02-23 08:04:03 +01:00
regnat
5687564a27 LocalStore: Send back the new realisations
To allow it to build ca derivations remotely
2021-02-23 08:04:03 +01:00
regnat
6fbf3fe636 Make the build-hook work with ca derivations
- Pass it the name of the outputs rather than their output paths (as
  these don't exist for ca derivations)
- Get the built output paths from the remote builder
- Register the new received realisations
2021-02-23 08:04:03 +01:00
Shea Levy
35205e2e92
Warn about instability of plugin API 2021-02-22 17:10:55 -05:00
Eelco Dolstra
8043bc6c8d
Merge pull request #4567 from NixOS/make-auto-call-error-eval-error
Make missing auto-call arguments throw an eval error
2021-02-22 16:35:52 +01:00
regnat
e2f3b2eb42 Make missing auto-call arguments throw an eval error
The PR #4240 changed messag of the error that was thrown when an auto-called
function was missing an argument.
However this change also changed the type of the error, from `EvalError`
to a new `MissingArgumentError`. This broke hydra which was relying on
an `EvalError` being thrown.

Make `MissingArgumentError` a subclass of `EvalError` to un-break hydra.
2021-02-22 16:13:09 +01:00
Eelco Dolstra
574eb2be81
Tweak error message 2021-02-22 15:24:14 +01:00
Daniël de Kok
2de232d2b3 Add x86_64 compute levels as additional system types
When performing distributed builds of machine learning packages, it
would be nice if builders without the required SIMD instructions can
be excluded as build nodes.

Since x86_64 has accumulated a large number of different instruction
set extensions, listing all possible extensions would be unwieldy.
AMD, Intel, Red Hat, and SUSE have recently defined four different
microarchitecture levels that are now part of the x86-64 psABI
supplement and will be used in glibc 2.33:

https://gitlab.com/x86-psABIs/x86-64-ABI
https://lwn.net/Articles/844831/

This change uses libcpuid to detect CPU features and then uses them to
add the supported x86_64 levels to the additional system types. For
example on a Ryzen 3700X:

$ ~/aps/bin/nix -vv --version | grep "Additional system"
Additional system types: i686-linux, x86_64-v1-linux, x86_64-v2-linux, x86_64-v3-linux
2021-02-22 09:11:15 +01:00
Domen Kožar
22aec8cef4
fix installer script 2021-02-21 15:51:49 +00:00
Domen Kožar
ae4260f0a7
Generate installer script for each PR/push
This works by using Cachix feature of serving a file from
a store path.
2021-02-21 15:51:49 +00:00
Eelco Dolstra
548437c234
Merge pull request #4541 from obsidiansystems/simpler-store-path-command
Deeper `Command` hierarchy to remove redundancy
2021-02-19 16:18:53 +01:00
regnat
f483b623e9 Remove the drv resolution caching mechanism
It isn't needed anymore now that don't need to eagerly resolve
everything like we used to do. So we can safely get rid of it
2021-02-19 15:48:31 +01:00
regnat
4bc28c44f2 Store the output hashes in the initialOutputs of the drv goal
That way we
1. Don't have to recompute them several times
2. Can compute them in a place where we know the type of the parent
  derivation, meaning that we don't need the casting dance we had before
2021-02-19 15:48:31 +01:00
Théophane Hufschmitt
0bfbd04369 Don't expose the "bang" drvoutput syntax
It's not fixed nor useful atm, so better keep it hidden

Co-authored-by: Eelco Dolstra <edolstra@gmail.com>
2021-02-19 15:48:31 +01:00
Théophane Hufschmitt
93d9eb78a0 Syntactic fixes
Co-authored-by: Eelco Dolstra <edolstra@gmail.com>
2021-02-19 15:48:31 +01:00
regnat
87c8d3d702 Register the realisations for unresolved drvs
Once a build is done, get back to the original derivation, and register
all the newly built outputs for this derivation.

This allows Nix to work properly with derivations that don't have all
their build inputs available − thus allowing garbage collection and
(once it's implemented) binary substitution
2021-02-19 15:48:31 +01:00
regnat
be1b5c4e59 Test the garbage collection of CA derivations
Simple test to ensure that `nix-build && nix-collect-garbage &&
nix-build -j0` works as it should
2021-02-19 15:48:31 +01:00
regnat
263f6dbd1c Don't crash nix-build when not all outputs are realised
Change the `nix-build` logic for linking/printing the output paths to allow for
some outputs to be missing. This might happen when the toplevel
derivation didn't have to be built, either because all the required
outputs were already there, or because they have all been substituted.
2021-02-19 15:48:31 +01:00
Eelco Dolstra
cd44c0af71 Increase default stack size on Linux
Workaround for #4550.
2021-02-18 19:22:37 +01:00
Eelco Dolstra
1b57825524 Document meta.mainProgram
Issue #4498.
2021-02-17 17:58:47 +01:00
Eelco Dolstra
7bd9898d5c nix run: Allow program name to be set in meta.mainProgram
This is useful when the program name doesn't match the package name
(e.g. ripgrep vs rg).

Fixes #4498.
2021-02-17 17:55:15 +01:00
Eelco Dolstra
13897afbe6 Throw an error if --arg / --argstr is used with a flake
Fixes #3949.
2021-02-17 17:32:10 +01:00
Eelco Dolstra
f33878b656 Make 'nix --version -vv' work
Fixes #3743.
2021-02-17 17:11:21 +01:00
Eelco Dolstra
cced73496b nix flake show: Handle 'overlays' output
Fixes #4542.
2021-02-17 16:53:39 +01:00
Eelco Dolstra
063de66909 nix develop: Fix quoted string handling
Fixes #4540.
2021-02-17 16:42:03 +01:00
Eelco Dolstra
6042febfce Restore warning about 'nix' being experimental
Fixes #4552.
2021-02-17 15:30:49 +01:00
Eelco Dolstra
54ab377288
Merge pull request #4553 from mausch/patch-1
Fix Haskell example
2021-02-17 10:18:42 +01:00
Mauricio Scheffer
5f4701e70d
Update doc/manual/src/command-ref/nix-shell.md
Co-authored-by: Cole Helbling <cole.e.helbling@outlook.com>
2021-02-16 23:27:04 +00:00
Mauricio Scheffer
35129884f9
Fix Haskell example
http://nixos.org redirects to https://nixos.org and apparently the HTTP library doesn't follow the redirect, so the output is empty.
When defining https in the request it crashes because the library doesn't seem to support https.
So this switches the example to a different http library.
2021-02-16 23:19:42 +00:00
John Ericson
ad337c8697 Deeper Command hierarchy to remove redundancy
Simply put, we now have `StorePathCommand : public StorePathsCommand` so
`StorePathCommand` doesn't reimplement work.
2021-02-12 17:48:09 +00:00
Eelco Dolstra
4e98f0345c
Merge pull request #4535 from NixOS/revert-4464-nar-narhash-addressed
Revert "narinfo: Change NAR URLs to be addressed on the NAR hash instead of the compressed hash"
2021-02-10 15:35:03 +01:00
Eelco Dolstra
2d4e1025f9
Merge pull request #4526 from NixOS/add-stale-bot
Add Stale bot
2021-02-10 11:44:27 +01:00
Graham Christensen
f2245091d0
Revert "narinfo: Change NAR URLs to be addressed on the NAR hash instead of the compressed hash" 2021-02-09 12:26:41 -05:00
Eelco Dolstra
ee3846b587
Merge pull request #4464 from tweag/nar-narhash-addressed
narinfo: Change NAR URLs to be addressed on the NAR hash instead of the compressed hash
2021-02-09 14:47:39 +01:00
Rok Garbas
bab3f30755
Auto closing issues/PRs after 1year. 2021-02-08 11:49:07 +01:00
Eelco Dolstra
fd6eaa1766
Merge pull request #4525 from sternenseemann/lowdown-0.8.0
libcmd/markdown: handle allocation errors in lowdown_term_rndr
2021-02-07 21:24:16 +01:00
Eelco Dolstra
37352aa7e1 Support --no-net for backwards compatibility 2021-02-07 20:44:56 +01:00
Alyssa Ross
7c112351d9
libutil: EPERM from kill(-1, ...) is fine
I tested a trivial program that called kill(-1, SIGKILL), which was
run as the only process for an unpriveleged user, on Linux and
FreeBSD.  On Linux, kill reported success, while on FreeBSD it failed
with EPERM.

POSIX says:

> If pid is -1, sig shall be sent to all processes (excluding an
> unspecified set of system processes) for which the process has
> permission to send that signal.

and

> The kill() function is successful if the process has permission to
> send sig to any of the processes specified by pid.  If kill() fails,
> no signal shall be sent.

and

> [EPERM]
>     The process does not have permission to send the signal to any
>     receiving process.

My reading of this is that kill(-1, ...) may fail with EPERM when
there are no other processes to kill (since the current process is
ignored).  Since kill(-1, ...) only attempts to kill processes the
user has permission to kill, it can't mean that we tried to do
something we didn't have permission to kill, so it should be fine to
interpret EPERM the same as success here for any POSIX-compliant
system.

This fixes an issue that Mic92 encountered[1] when he tried to review a
Nixpkgs PR on FreeBSD.

[1]: https://github.com/NixOS/nixpkgs/pull/81459#issuecomment-606073668
2021-02-07 13:56:50 +00:00
Rok Garbas
91d83426f7
typo 2021-02-06 13:33:34 +01:00
Rok Garbas
6af26b7aec
Add Stale bot
The configuration was taken from nixpkgs repository and adjusted to
`NixOS/nix`.

A `stale` label was added to the labels (with gray color).

Issues and PRs with `critical` label are excluded from interacting with the
stale bot.
2021-02-06 13:29:38 +01:00
sternenseemann
d0e34c85f8 libcmd/markdown: handle allocation errors in lowdown_term_rndr
We upgrade to lowdown 0.8.0 [1] which contains a fix/improvement to a
behavior mentioned in this issue thread [2] where a big part of
lowdown's API would just call exit(1) on allocation errors since that
is a satisfying behavior for the lowdown binary.

Now lowdown_term_rndr returns 0 if an allocation error occurred which we
check for in libcmd/markdown.cc.

Also the extern "C" { } wrapper around lowdown.h has been removed as it
is not necessary.

[1]: 6ca7c855a0/versions.xml (L987-L1006)
[2]: https://github.com/kristapsdz/lowdown/issues/45#issuecomment-756681153
2021-02-06 13:14:57 +01:00
Eelco Dolstra
480426a364 Add more instrumentation for #4270 2021-02-05 15:57:33 +01:00
Eelco Dolstra
d7c27f21ab
Merge pull request #4372 from tweag/ca/drvoutputs-commands
Add a new Cmd type working on RealisedPaths
2021-02-05 13:03:50 +01:00