On Tue, May 3, 2016 at 3:26 PM, Julian Seward <[email protected]> wrote: > >> Inspired by this, I've been thinking about introducing a "verification >> service". > > Ah, now there's an interesting idea. Ideally you'd want some kind of > low overhead continuous coverage, at least for m-c/aurora/beta. I'm > imagining an always-on service that periodically sends verify-yourself > requests to subsystems and aggregates the results. It could possibly > learn the approximate time required by each subsystem to verify, so that > it can dynamically adjust the request frequencies so as to achieve a global > cost under some limit, eg 1% total CPU time.
I had only considered explicit triggering, but periodic checks in the wild would greatly improve coverage. > Perhaps even use memory reporter info to drive the scheduling, on the > assumption that the time to verify subsystem X is proportional to the > amount of data that subsystem X has. I'm not sure that assumption holds. Another possibility would be to do a similar thing to telemetry, which I think happens only after the browser's been idle for a while, or something. > I presume you're thinking about all of Gecko now, and not just SM? I am. Yes! Nick _______________________________________________ dev-tech-js-engine-internals mailing list [email protected] https://lists.mozilla.org/listinfo/dev-tech-js-engine-internals
