qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Markus Armbruster <armbru@redhat.com>
To: Paolo Bonzini <pbonzini@redhat.com>
Cc: qemu-devel@nongnu.org, marcandre.lureau@redhat.com,
	qemu-rust@nongnu.org,  John Snow <jsnow@redhat.com>
Subject: Re: [PATCH 13/19] scripts/qapi: add QAPISchemaIfCond.rsgen()
Date: Tue, 09 Dec 2025 19:43:48 +0100	[thread overview]
Message-ID: <87ms3rbimz.fsf@pond.sub.org> (raw)
In-Reply-To: <20251010151006.791038-14-pbonzini@redhat.com> (Paolo Bonzini's message of "Fri, 10 Oct 2025 17:09:58 +0200")

Paolo Bonzini <pbonzini@redhat.com> writes:

> From: Marc-André Lureau <marcandre.lureau@redhat.com>
>
> Generate Rust #[cfg(...)] guards from QAPI 'if' conditions.

Please mention that this isn't used, yet.  I commonly write something
like "The next commit will put it to use."

> Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
> Link: https://lore.kernel.org/r/20210907121943.3498701-15-marcandre.lureau@redhat.com
> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
> ---
>  scripts/qapi/common.py | 16 ++++++++++++++++
>  scripts/qapi/schema.py |  4 ++++
>  2 files changed, 20 insertions(+)
>
> diff --git a/scripts/qapi/common.py b/scripts/qapi/common.py
> index d7c8aa3365c..f16b9568bb9 100644
> --- a/scripts/qapi/common.py
> +++ b/scripts/qapi/common.py
> @@ -199,6 +199,22 @@ def guardend(name: str) -> str:
>                   name=c_fname(name).upper())
>  
>  
> +def rsgen_ifcond(ifcond: Optional[Union[str, Dict[str, Any]]]) -> str:

This is the Rust-generating cousin of cgen_ifcond().

The argument is None or a tree.  The tree's leaves are str, and its
inner nodes look like

    {'all': [sub-tree, ...]}
    {'any': [sub-tree, ...]}
    {'not': sub-tree}

mypy doesn't do recursive types, so we approximate the tree as
Union[str, Dict[str, Any]].

> +
> +    def cfg(ifcond: Union[str, Dict[str, Any]]) -> str:

The argument's type is wrong.  It should be

    Union[str, List[Dict[str, Any]], Dict[str, Any]]

We'll see below why mypy doesn't complain, and where the List[...]
comes from.

The name @ifcond is misleading.  This isn't an if condition, it's the
union of if condition and list of if conditions.

I needed John Snow's help to figure this out.  Thanks, John!

Case 1: the code for str, i.e. a tree leaf:

> +        if isinstance(ifcond, str):
> +            return ifcond

Case 2: the code for List[Dict[str, Any]]:

> +        if isinstance(ifcond, list):
> +            return ', '.join([cfg(c) for c in ifcond])

Case 3: the code for Dict[str, Any]:

gen_ifcond() below has

           assert isinstance(ifcond, dict) and len(ifcond) == 1

right here to make this crystal clear.

> +        oper, operands = next(iter(ifcond.items()))

Recall @ifcond is either

    {'all': [sub-tree, ...]}
    {'any': [sub-tree, ...]}
    {'not': sub-tree}

The next(...) wizardry returns the first element of the @ifcond
dictionary.  Actually *the* element, since @ifcond has just one.

Thus:

    @oper is 'all', 'any', or 'not'

    @operands is a sub-tree when @oper is 'not', else a [sub-tree, ...],
    i.e. Dict[str, Any] or List[Dict[str, Any]]

> +        operands = cfg(operands)

We pass @operands to cfg().  That's where the List[...] comes from.

> +        return f'{oper}({operands})'
> +
> +    if not ifcond:
> +        return ''
> +    return '#[cfg(%s)]' % cfg(ifcond)

So, cfg(ifcond) returns the argument to interpolate into '#[cfg(%s)]'.

When @ifcond is str, it's @ifcond itself.  This is case 1.

When @ifcond is {'not': COND}, it's 'not(CC)', where CC is cfg(COND).
This is case 3 and case 2 with a non-list argument.

When @ifcond is {'all': [COND, ...]}, it's 'all(CC, ...)', where the CC
are cfg(COND).  This is case 3 and case 2 with a list argument.

Likewise for {'any': [COND, ...]}.

Okay apart from the incorrect type hint and the misleading name.

Less clever code would've saved me quite some review time.

But why doesn't mypy scream?  Consider again

           oper, operands = next(iter(ifcond.items()))

@ifcond's static type is Dict[str, Any].  Therefore @oper's static type
is str, and @operands is Any.  Any suppresses type checking!  The call
cfg(operands) is therefore *not* checked, and we get away with passing a
list to cfg() even though its type hint doesn't allow it.

> +
> +
>  def gen_ifcond(ifcond: Optional[Union[str, Dict[str, Any]]],
>                 cond_fmt: str, not_fmt: str,
>                 all_operator: str, any_operator: str) -> str:
> diff --git a/scripts/qapi/schema.py b/scripts/qapi/schema.py
> index 8d88b40de2e..848a7401251 100644
> --- a/scripts/qapi/schema.py
> +++ b/scripts/qapi/schema.py
> @@ -37,6 +37,7 @@
>      docgen_ifcond,
>      gen_endif,
>      gen_if,
> +    rsgen_ifcond,
>  )
>  from .error import QAPIError, QAPISemError, QAPISourceError
>  from .expr import check_exprs
> @@ -63,6 +64,9 @@ def gen_endif(self) -> str:
>      def docgen(self) -> str:
>          return docgen_ifcond(self.ifcond)
>  
> +    def rsgen(self) -> str:
> +        return rsgen_ifcond(self.ifcond)
> +
>      def is_present(self) -> bool:
>          return bool(self.ifcond)



  reply	other threads:[~2025-12-09 18:44 UTC|newest]

Thread overview: 48+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-10-10 15:09 [PATCH 00/19] rust: QObject and QAPI bindings Paolo Bonzini
2025-10-10 15:09 ` [PATCH 01/19] util: add ensure macro Paolo Bonzini
2025-10-10 15:09 ` [PATCH 02/19] rust/util: use anyhow's native chaining capabilities Paolo Bonzini
2025-10-10 15:09 ` [PATCH 03/19] rust: do not add qemuutil to Rust crates Paolo Bonzini
2025-12-05  8:30   ` Markus Armbruster
2025-12-05 12:07     ` Paolo Bonzini
2025-10-10 15:09 ` [PATCH 04/19] rust/qobject: add basic bindings Paolo Bonzini
2025-12-05  9:35   ` Markus Armbruster
2025-12-05 11:27     ` Paolo Bonzini
2025-12-11  7:21       ` Markus Armbruster
2025-10-10 15:09 ` [PATCH 05/19] subprojects: add serde Paolo Bonzini
2025-10-10 15:09 ` [PATCH 06/19] rust/qobject: add Serialize implementation Paolo Bonzini
2025-12-05  9:47   ` Markus Armbruster
2025-12-10 17:19     ` Paolo Bonzini
2025-12-11  7:07       ` Markus Armbruster
2025-10-10 15:09 ` [PATCH 07/19] rust/qobject: add Serializer (to_qobject) implementation Paolo Bonzini
2025-10-10 15:09 ` [PATCH 08/19] rust/qobject: add Deserialize implementation Paolo Bonzini
2025-10-10 15:09 ` [PATCH 09/19] rust/qobject: add Deserializer (from_qobject) implementation Paolo Bonzini
2025-10-10 15:09 ` [PATCH 10/19] rust/util: replace Error::err_or_unit/err_or_else with Error::with_errp Paolo Bonzini
2025-10-10 15:09 ` [PATCH 11/19] rust/qobject: add from/to JSON bindings for QObject Paolo Bonzini
2025-12-05 10:04   ` Markus Armbruster
2025-12-05 11:09     ` Paolo Bonzini
2025-12-05 12:16       ` Markus Armbruster
2025-12-08  7:00         ` Paolo Bonzini
2025-12-08  9:17           ` Markus Armbruster
2025-12-09  7:34             ` Paolo Bonzini
2025-10-10 15:09 ` [PATCH 12/19] rust/qobject: add Display/Debug Paolo Bonzini
2025-10-10 15:09 ` [PATCH 13/19] scripts/qapi: add QAPISchemaIfCond.rsgen() Paolo Bonzini
2025-12-09 18:43   ` Markus Armbruster [this message]
2025-10-10 15:09 ` [PATCH 14/19] scripts/qapi: generate high-level Rust bindings Paolo Bonzini
2025-12-09 10:03   ` Markus Armbruster
2025-12-10 14:38     ` Paolo Bonzini
2025-12-11  7:25       ` Markus Armbruster
2025-10-10 15:10 ` [PATCH 15/19] scripts/qapi: add serde attributes Paolo Bonzini
2025-12-09 12:24   ` Markus Armbruster
2025-10-10 15:10 ` [PATCH 16/19] scripts/qapi: strip trailing whitespaces Paolo Bonzini
2025-12-09  8:48   ` Markus Armbruster
2025-12-10 17:28     ` Paolo Bonzini
2025-10-10 15:10 ` [PATCH 17/19] scripts/rustc_args: add --no-strict-cfg Paolo Bonzini
2025-10-10 15:10 ` [PATCH 18/19] rust/util: build QAPI types Paolo Bonzini
2025-10-10 15:10 ` [PATCH 19/19] rust/tests: QAPI integration tests Paolo Bonzini
2025-10-30 17:13 ` [PATCH 00/19] rust: QObject and QAPI bindings Paolo Bonzini
2025-12-05 13:55 ` Markus Armbruster
2025-12-09  6:01   ` Markus Armbruster
2025-12-10 14:12     ` Paolo Bonzini
2025-12-10 14:50       ` Markus Armbruster
2025-12-10 16:01         ` Paolo Bonzini
2025-12-10 16:59           ` Markus Armbruster

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=87ms3rbimz.fsf@pond.sub.org \
    --to=armbru@redhat.com \
    --cc=jsnow@redhat.com \
    --cc=marcandre.lureau@redhat.com \
    --cc=pbonzini@redhat.com \
    --cc=qemu-devel@nongnu.org \
    --cc=qemu-rust@nongnu.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).