Hi Ms2ger! Can you briefly summarize the new approach, contrast it to the old one, and describe how it solves the issues around moving?
Thanks! Nick On Mon, Jun 27, 2016 at 1:21 AM, Ms2ger <ms2...@gmail.com> wrote: > [Cross-posting to m.d.t.j-e.internals; please reply only to m.d.servo.] > > Hi all, > > eddyb is proposing a new, macro-based approach to rooting SpiderMonkey > types (i.e. types that use `js::jsapi::Rooted`, NOT Servo DOM objects, > which still use `script::dom::bindings::js::Root`), as well as the > fairly uncommon `script::dom::bindings::trace::RootedVec`. > > The goal is to stop using the `return_address` intrinsic, which is > fundamentally incompatible with Rust's assumption that every type can be > moved silently. We have been lucky enough to have the `return_address` > approach work so far, but the introduction of MIR (mid-level > intermediate representation; a large-scale refactoring of code > generation within the Rust compiler) will break our code when it is > turned on. > > Typical use changes from > > ``` > let obj = RootedObject::new(cx, self.callback()); > ``` > > to > > ``` > rooted!(in(cx) let obj = self.callback()); > ``` > > The pull requests are at: > > https://github.com/servo/rust-mozjs/pull/272 > https://github.com/servo/servo/pull/11872 > > Comments from SpiderMonkey GC hackers are or course very welcome. > > Ms2ger > _______________________________________________ > dev-servo mailing list > dev-servo@lists.mozilla.org > https://lists.mozilla.org/listinfo/dev-servo > _______________________________________________ dev-servo mailing list dev-servo@lists.mozilla.org https://lists.mozilla.org/listinfo/dev-servo
