Stream: project leads (public)

Topic: catch_unwind is a problem


simulacrum (Dec 24 2019 at 00:45, on Zulip):

@James Munns (Embedded), @Josh Triplett (FFI), and @Alex Crichton (WASM) -- we're trying to resolve in #64224 whether it's worth trying to bring the code size of catch_unwind down even further. It sounds like at least Servo has no data showing that the performance improvement such a change could bring is worth it, but I'm wondering if any of you are aware if your domains would benefit from help here.

If you have thoughts here I'd appreciate a comment on that issue or I'm happy to discuss here as well!

simulacrum (Dec 24 2019 at 00:45, on Zulip):

cc @nikomatsakis as well, I'm not sure who is leading the FFI unwinding effort but they might be interested in this as well

Alex Crichton (Dec 24 2019 at 01:29, on Zulip):

I will leave a comment when back from break

XAMPPRocky (Dec 24 2019 at 16:26, on Zulip):

@simulacrum I believe @Kyle Strand is leading ffi unwind with niko.

Kyle Strand (Dec 24 2019 at 18:08, on Zulip):

Also @acfoltzer

simulacrum (Dec 26 2019 at 23:51, on Zulip):

As a bit of a follow up we've since ended up investing some time with @Amanieu and should have reached parity with C++.

nikomatsakis (Dec 30 2019 at 20:41, on Zulip):

Where can I learn more about the follow-up work that was done?

Amanieu (Dec 30 2019 at 21:22, on Zulip):

@nikomatsakis https://github.com/rust-lang/rust/pull/67502

Alex Crichton (Jan 06 2020 at 15:06, on Zulip):

To follow up from this I don't think catch_unwind affects wasm at all right now since wasm is always panic=abort

Last update: Jul 03 2020 at 17:05UTC