On 9/28/21 10:30, Raffaello Giulietti wrote: > I was pondering whether to propose adding a launch-time option on the command > line to choose between the current and the proposed implementation for some > time, so to help the community gaining confidence in the new algorithm and > still have the current one as a fallback, just in case. (AFAIU, a launch-time > option can be constant folded by the JIT compiler to help it eliminate dead > code, right?) > > On the one hand, this adds complexity. On the other hand, it could help > revive this dormant PR. > > What do you think? Would this be a viable option?
It's a good-enough idea in theory, but I don't think it'd be accepted. Pinging Joe Darcy: do you (or any of your friends) have a good way to contact Guy Steele? -- Andrew Haley (he/him) Java Platform Lead Engineer Red Hat UK Ltd. <https://www.redhat.com> https://keybase.io/andrewhaley EAC8 43EB D3EF DB98 CC77 2FAD A5CD 6035 332F A671
