[PATCH 0/3] scripts/container: Minor fixups suggested by ruff

[PATCH 0/3] scripts/container: Minor fixups suggested by ruff

[Nathan Chancellor]

Hi all,

This series fixes a few warnings that I see when running

  $ ruff check --select C4,RUF scripts/container

which were the few warnings from my personal ruff.toml that seemed most
interesting.

I will apply these on top of the initial scripts/container change.

---
Nathan Chancellor (3):
      scripts/container: Turn runtimes class variable into a tuple
      scripts/container: Use list comprehension in get_names()
      scripts/container: Use iterable unpacking for _get_opts()

 scripts/container | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)
---
base-commit: 6eac13c876805f61bbb588eaff5ada0b6dc603e8
change-id: 20260122-scripts-container-ruff-fixes-4d38fc92771a

Best regards,
--  
Nathan Chancellor <nathan@kernel.org>

[PATCH 1/3] scripts/container: Turn runtimes class variable into a tuple

[Nathan Chancellor]

'ruff check --select RUF scripts/container' warns:

  RUF012 Mutable class attributes should be annotated with `typing.ClassVar`
     --> scripts/container:108:16
      |
  106 |     """List of all supported runtimes"""
  107 |
  108 |     runtimes = [PodmanRuntime, DockerRuntime]
      |                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  109 |
  110 |     @classmethod
      |

runtimes is never modified so move from a list to a tuple, clearing up
the warning.

Signed-off-by: Nathan Chancellor <nathan@kernel.org>
---
 scripts/container | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/scripts/container b/scripts/container
index b05333d8530b..dae475c6d262 100755
--- a/scripts/container
+++ b/scripts/container
@@ -105,7 +105,7 @@ class PodmanRuntime(CommonRuntime):
 class Runtimes:
     """List of all supported runtimes"""
 
-    runtimes = [PodmanRuntime, DockerRuntime]
+    runtimes = (PodmanRuntime, DockerRuntime)
 
     @classmethod
     def get_names(cls):

-- 
2.52.0

Re: [PATCH 1/3] scripts/container: Turn runtimes class variable into a tuple

[Guillaume Tucker]

On 23/01/2026 12:27 am, Nathan Chancellor wrote:
> -    runtimes = [PodmanRuntime, DockerRuntime]
> +    runtimes = (PodmanRuntime, DockerRuntime)

My 2¢ - A subtle difference between a non-mutable list and a tuple,
is that a list is a series of values with arbitrary length whereas a
tuple is a single variable with a fixed number of values meant to be
used together.  Say, coordinates make sense as a tuple but not as a
list.  Search results from a database make sense to be stored in a
list but not as a tuple.  In this case, there could be any number of
runtimes so it's semantically a list.  At least that's why I used a
list, now there may be good reasons for using a tuple too.

Cheers,
Guillaume

[PATCH 2/3] scripts/container: Use list comprehension in get_names()

[Nathan Chancellor]

'ruff check --select C4 scripts/container' warns:

  C400 Unnecessary generator (rewrite as a list comprehension)
     --> scripts/container:113:16
      |
  111 |     def get_names(cls):
  112 |         """Get a list of all the runtime names"""
  113 |         return list(runtime.name for runtime in cls.runtimes)
      |                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  114 |
  115 |     @classmethod
      |
  help: Rewrite as a list comprehension

Do as it suggests.

Signed-off-by: Nathan Chancellor <nathan@kernel.org>
---
 scripts/container | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/scripts/container b/scripts/container
index dae475c6d262..5afaad762492 100755
--- a/scripts/container
+++ b/scripts/container
@@ -110,7 +110,7 @@ class Runtimes:
     @classmethod
     def get_names(cls):
         """Get a list of all the runtime names"""
-        return list(runtime.name for runtime in cls.runtimes)
+        return [runtime.name for runtime in cls.runtimes]
 
     @classmethod
     def get(cls, name):

-- 
2.52.0

Re: [PATCH 2/3] scripts/container: Use list comprehension in get_names()

[Guillaume Tucker]

On 23/01/2026 12:27 am, Nathan Chancellor wrote:
> -        return list(runtime.name for runtime in cls.runtimes)
> +        return [runtime.name for runtime in cls.runtimes]

My 2¢ - A generator is a generalisation of list comprehension, so
using [] is syntactic sugar for list().  I prefer to always spell out
the type explicitly with generators to avoid bad surprises with {}
which could be seen either as a set() or a dict().

Cheers,
Guillaume

[PATCH 3/3] scripts/container: Use iterable unpacking for _get_opts()

[Nathan Chancellor]

'ruff check --select RUF scripts/container' warns:

  RUF005 Consider iterable unpacking instead of concatenation
    --> scripts/container:89:16
     |
  88 |       def _get_opts(self, container_name):
  89 |           return super()._get_opts(container_name) + [
     |  ________________^
  90 | |             '--user', f'{self._uid}:{self._gid}'
  91 | |         ]
     | |_________^
     |
  help: Replace with iterable unpacking

  RUF005 Consider iterable unpacking instead of concatenation
     --> scripts/container:100:16
      |
   99 |       def _get_opts(self, container_name):
  100 |           return super()._get_opts(container_name) + [
      |  ________________^
  101 | |             '--userns', f'keep-id:uid={self._uid},gid={self._gid}',
  102 | |         ]
      | |_________^
      |
  help: Replace with iterable unpacking

Iterable unpacking is more efficient, so switch to it as it suggests
(even if it is unlikely to matter given how small these lists are).

Signed-off-by: Nathan Chancellor <nathan@kernel.org>
---
 scripts/container | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/scripts/container b/scripts/container
index 5afaad762492..09a14f7e29f3 100755
--- a/scripts/container
+++ b/scripts/container
@@ -86,7 +86,8 @@ class DockerRuntime(CommonRuntime):
     name = 'docker'
 
     def _get_opts(self, container_name):
-        return super()._get_opts(container_name) + [
+        return [
+            *super()._get_opts(container_name),
             '--user', f'{self._uid}:{self._gid}'
         ]
 
@@ -97,7 +98,8 @@ class PodmanRuntime(CommonRuntime):
     name = 'podman'
 
     def _get_opts(self, container_name):
-        return super()._get_opts(container_name) + [
+        return [
+            *super()._get_opts(container_name),
             '--userns', f'keep-id:uid={self._uid},gid={self._gid}',
         ]
 

-- 
2.52.0

Re: [PATCH 3/3] scripts/container: Use iterable unpacking for _get_opts()

[Guillaume Tucker]

On 23/01/2026 12:27 am, Nathan Chancellor wrote:
> -        return super()._get_opts(container_name) + [
> +        return [
> +            *super()._get_opts(container_name),
>               '--user', f'{self._uid}:{self._gid}'
>           ]

My last 2¢ - Yes it's more efficient to build the list in one go but
I find it easier to read and maintain with the + [...options...]
syntax.  I guess it's hard to tell a linter that sometimes, runtime
performance isn't so important.

These are all minor things but they come with interesting language
implications so I thought it was worth a few comments...

Cheers,
Guillaume

Re: [PATCH 0/3] scripts/container: Minor fixups suggested by ruff

[Miguel Ojeda]

On Fri, Jan 23, 2026 at 12:27 AM Nathan Chancellor <nathan@kernel.org> wrote:
>
> This series fixes a few warnings that I see when running
>
>   $ ruff check --select C4,RUF scripts/container
>
> which were the few warnings from my personal ruff.toml that seemed most
> interesting.

I haven't had time to look into the new container support yet, but
having kernel Python scripts Ruff-clean sounds good in general, and I
wonder -- should we consider having a `.ruff.toml` eventually?

At least we have a small `.pylintrc` already, which is quite recent by the way.

I think I recall you also using Black and/or Flake8, for
ClangBuiltLinux-related bits perhaps?

Thanks!

Cheers,
Miguel

Re: [PATCH 0/3] scripts/container: Minor fixups suggested by ruff

[Nathan Chancellor]

On Fri, Jan 23, 2026 at 04:14:16PM +0100, Miguel Ojeda wrote:
> On Fri, Jan 23, 2026 at 12:27 AM Nathan Chancellor <nathan@kernel.org> wrote:
> >
> > This series fixes a few warnings that I see when running
> >
> >   $ ruff check --select C4,RUF scripts/container
> >
> > which were the few warnings from my personal ruff.toml that seemed most
> > interesting.
> 
> I haven't had time to look into the new container support yet, but
> having kernel Python scripts Ruff-clean sounds good in general, and I
> wonder -- should we consider having a `.ruff.toml` eventually?

I think that we should. Even something as simple as the one in the Ruff
docs would keep things consistent:

  https://docs.astral.sh/ruff/linter/#rule-selection

I guess the one problem there is consistency amongst developers (since
one's linting workflow is fairly personal) and having a wide range of
Python code across the tree but having a .ruff.toml for people to run if
they want it is harmless.

> At least we have a small `.pylintrc` already, which is quite recent by the way.

Ah, good to know!

> I think I recall you also using Black and/or Flake8, for
> ClangBuiltLinux-related bits perhaps?

We use Ruff and pylint for linting, yapf for formatting. Ruff can
replace flake8 outright with the right configuration in my experience.

Cheers,
Nathan

Re: [PATCH 0/3] scripts/container: Minor fixups suggested by ruff

[Miguel Ojeda]

On Fri, Jan 23, 2026 at 10:01 PM Nathan Chancellor <nathan@kernel.org> wrote:
>
> I think that we should. Even something as simple as the one in the Ruff
> docs would keep things consistent:
>
>   https://docs.astral.sh/ruff/linter/#rule-selection
>
> I guess the one problem there is consistency amongst developers (since
> one's linting workflow is fairly personal) and having a wide range of
> Python code across the tree but having a .ruff.toml for people to run if
> they want it is harmless.

Yeah, I think as long as it is optional (and the rules reasonable --
not sure how opinionated the defaults are, but we can tweak things as
needed), then it is a win with little downside.

It is similar to the clang-format situation -- some developers take
advantage of it, others don't.

Over time, the idea would be to try to define the set of rules that
most people is happy with, and thus to get more and more files to
conform.

Thanks!

Cheers,
Miguel

Re: [PATCH 0/3] scripts/container: Minor fixups suggested by ruff

[Guillaume Tucker]

Hi Nathan,

On 23/01/2026 12:27 am, Nathan Chancellor wrote:
> Hi all,
> 
> This series fixes a few warnings that I see when running
> 
>    $ ruff check --select C4,RUF scripts/container
> 
> which were the few warnings from my personal ruff.toml that seemed most
> interesting.
> 
> I will apply these on top of the initial scripts/container change.
> 
> ---
> Nathan Chancellor (3):
>        scripts/container: Turn runtimes class variable into a tuple
>        scripts/container: Use list comprehension in get_names()
>        scripts/container: Use iterable unpacking for _get_opts()
> 
>   scripts/container | 10 ++++++----
>   1 file changed, 6 insertions(+), 4 deletions(-)
> ---
> base-commit: 6eac13c876805f61bbb588eaff5ada0b6dc603e8
> change-id: 20260122-scripts-container-ruff-fixes-4d38fc92771a

LGTM - some of these things are arguably more a matter of individual
taste than absolute best practices but it's good to have a
well-defined code quality standard.  I typically use Pylint,
Pycodestyle and Mypy, not Ruff or Black.  So for the whole series:

Reviewed-by: Guillaume Tucker <gtucker@gtucker.io>

Cheers,
Guillaume

Re: [PATCH 0/3] scripts/container: Minor fixups suggested by ruff

[Nathan Chancellor]

On Fri, Jan 23, 2026 at 09:26:20PM +0100, Guillaume Tucker wrote:
> Hi Nathan,
> 
> On 23/01/2026 12:27 am, Nathan Chancellor wrote:
> > Hi all,
> > 
> > This series fixes a few warnings that I see when running
> > 
> >    $ ruff check --select C4,RUF scripts/container
> > 
> > which were the few warnings from my personal ruff.toml that seemed most
> > interesting.
> > 
> > I will apply these on top of the initial scripts/container change.
> > 
> > ---
> > Nathan Chancellor (3):
> >        scripts/container: Turn runtimes class variable into a tuple
> >        scripts/container: Use list comprehension in get_names()
> >        scripts/container: Use iterable unpacking for _get_opts()
> > 
> >   scripts/container | 10 ++++++----
> >   1 file changed, 6 insertions(+), 4 deletions(-)
> > ---
> > base-commit: 6eac13c876805f61bbb588eaff5ada0b6dc603e8
> > change-id: 20260122-scripts-container-ruff-fixes-4d38fc92771a
> 
> LGTM - some of these things are arguably more a matter of individual
> taste than absolute best practices but it's good to have a
> well-defined code quality standard.  I typically use Pylint,
> Pycodestyle and Mypy, not Ruff or Black.  So for the whole series:
> 
> Reviewed-by: Guillaume Tucker <gtucker@gtucker.io>

Thanks a lot for taking a look and providing the comments. I am going to
hold off on this series for a little bit to ponder on providing a
.ruff.toml file first to ensure we have a common understanding around
Python linting in the kernel, especially given a lot of these lints can
be opinionated as you mentioned.

Cheers,
Nathan

Re: [PATCH 0/3] scripts/container: Minor fixups suggested by ruff

[Masahiro Yamada]

On Fri, Jan 23, 2026 at 8:27 AM Nathan Chancellor <nathan@kernel.org> wrote:
>
> Hi all,
>
> This series fixes a few warnings that I see when running
>
>   $ ruff check --select C4,RUF scripts/container
>
> which were the few warnings from my personal ruff.toml that seemed most
> interesting.
>
> I will apply these on top of the initial scripts/container change.
>
> ---
> Nathan Chancellor (3):
>       scripts/container: Turn runtimes class variable into a tuple
>       scripts/container: Use list comprehension in get_names()
>       scripts/container: Use iterable unpacking for _get_opts()
>
>  scripts/container | 10 ++++++----
>  1 file changed, 6 insertions(+), 4 deletions(-)


scripts/container is written in a generic way.

In my understanding, this runs an arbitrary
command in a container, and that's it.
Nothing is related to Kbuild.

$ scripts/container  -i debian:trixie  echo hello
hello

If this script is really useful,
why don't you keep this outside the kernel code
so that non-kernel developers can use it?


-- 
Best Regards
Masahiro Yamada