Since withFramedSink() is now used a lot more than in the past (for
every addToStore() variant), we were creating a lot of threads, e.g.
nix flake show --no-eval-cache --all-systems github:NixOS/nix/afdd12be5e19c0001ff3297dea544301108d298
would create 46418 threads. While threads on Linux are cheap, this is
still substantial overhead.
So instead, just poll from FramedSink before every write whether there
are pending messages from the daemon. This could slightly increase the
latency on log messages from the daemon, but not on exceptions (which
were only synchronously checked from FramedSink anyway).
This speeds up the command above from 19.2s to 17.5s on my machine (a
9% speedup).
Fixes
```
umount: /tmp/nix-shell.i3xRwX/nix-test/local-overlay-store/delete-refs/stores/merged-store/nix/store: filesystem was unmounted, but failed to update userspace mount table.
make: *** [mk/lib.mk:93: tests/functional/local-overlay-store/delete-refs.sh.test] Error 16
```
in a dev shell.
Note: this previously worked before we didn't have umount in the dev
shell, so we got /run/wrappers/bin/umount.
Incorrectly high expectations lead to frustration for users who
stick around to experience how useless it is for e.g. a devShell
https://functional.cafe/@arianvp/112976284363120036:
> Flakes doesn't have eval caching. It has command line argument
> caching. It literally just stores the cli argument you passed
> in a sqlite database and yes that's as useless as it sounds
> When I discovered flakes had no expression level caching whatsoever
> I kind of felt lied to and betrayed.
Fixes
```
GEN /home/eelco/Dev/nix-master/outputs/out/share/doc/nix/manual/index.html
error: File not found: ../store/types/
┌─ release-notes/rl-next.md:60:197
│
60 │ The build hook protocol did in principle support custom ways of remote building, but that can also be accomplished with a custom service for the ssh or daemon/ssh-ng protocols, or with a custom [store type](../store/types/) i.e. `Store` subclass. <!-- we normally don't mention classes, but consider that this release note is about a library use case -->
│ ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File not found: ../store/types/
Error: One or more incorrect links
2024-08-19 16:47:57 [ERROR] (mdbook::renderer): Renderer exited with non-zero return code.
2024-08-19 16:47:57 [ERROR] (mdbook::utils): Error: Rendering failed
2024-08-19 16:47:57 [ERROR] (mdbook::utils): Caused By: The "linkcheck" renderer failed
```
`make check` was reverted too soon. The hacking guide wasn't brought
up to date with the new workflow, and it's not clear how to use
meson for everything.
This reverts commit 6f3045c2a2.
The current backport action cannot automerge because
the github action bot does not trigger github CI actions.
Mergify instead does not have this limitation and can also
use a merge queue.
On top we have now a declarative configuration to allow
contributers to add new tests to required without having access
to the github org.
An example pull request and backport can be seen here:
https://github.com/Mic92/nix-1/pull/4
and here:
https://github.com/Mic92/nix-1/pull/5
To complete the setup the mergify app must be enabled for this repository.
It's already installed in the nixos organization for nixos-hardware and
other repositories.
In d60c3f7f7c, this was changed to close a
hole in the sandbox. Unfortunately, this was too restrictive such that it
made local port binding fail, thus making derivations that needed
`__darwinAllowLocalNetworking` gain nearly nothing, and thus largely
fail (as the primary use for it is to enable port binding).
This unfortunately does mean that a sandboxed build process can, in
coordination with an actor outside the sandbox, escape the sandbox by
binding a port and connecting to it externally to send data. I do not
see a way around this with my experimentation and understanding of the
(quite undocumented) macOS sandbox profile API. Notably it seems not
possible to use the sandbox to do any of:
- Restrict the remote IP of inbound network requests
- Restrict the address being bound to
As such, the `(local ip "*:*")` here appears to be functionally no
different than `(local ip "localhost:*")` (however it *should* be
different than removing the filter entirely, as that would make it also
apply to non-IP networking). Doing `(allow network-inbound (require-all
(local ip "localhost:*") (remote ip "localhost:*")))` causes listening
to fail.
Note that `network-inbound` implies `network-bind`.
(cherry picked from commit 00f6db36fd)
This wasn't the default behaviour because:
> We don't enable this by default to avoid the mostly unnecessary work of
> performing an additional build of the package in cases where we build
> the package normally anyway, such as in our pre-merge CI.
Since we have a componentized build, we've solved the duplication.
In the new situation, building both with and without unit tests
isn't any slow than just a build with unit tests, so there's no
point in using the unit-tested build anymore.
By using the otherwise untested build, we reduce the minimum build
time towards the NixOS test, at no cost.
If you want to run all tests, build all attributes.
Setting it to /bin/sh will make it more predictable when users have
their favorite shell in SHELL, which might not behave as expected.
For instance, a bad rc file could send something to stdout before
our LocalCommand gets to write "started".
This may help https://github.com/NixOS/nix/issues/11010
