On Fri, Jun 13, 2025 at 7:36 AM Zhao Liu <zhao1....@intel.com> wrote: > > 1) everything is done through an intermediate QObject step (e.g. for a > > setter: Visitor->QObject with an input visitor, and QObject->Rust with > > serde deserialization). > > + easy, Rust only sees serde > > + QMP commands use a single conversion step > > - inefficient > > This sounds like a natural approach.
Yep, serde is complex but the result is very natural. In some cases serde has different defaults than QEMU, but attributes like #[serde(skip_serializing_if = "Option::is_none")] can be automatically generated by qapi-gen. > > 2) everything is done through an intermediate C step (e.g. for a > > setter: Visitor->C with a visit_type_* function, and C->Rust with > > generated code that does not need to use serde). > > I understand this step indicates to use bindgen to generate visit_type_* > bindings... More than that, it's using qapi-gen to generate implementations like the *Foreign traits used by Error. Marc-André wrote his own traits but I suspect a lot of the code could be reused. So I must give credit to Marc-André that there is a big advantage here, in that a lot of the code is already written and it is simpler than serde. Paolo
