Potential type instability with prospective 1.6 type inference changes
lassepe opened this issue · comments
There is some potential type instability coming our way:
JuliaLang/julia#36208 (comment)
This is probably one more reason to register all solvers in the general registry so that all of them are covered by PkgEval reports.
Wow, you did not waste any time noticing this! :) But stay tuned, I might have a mitigation in Base for this.
Haha wow @lassepe , how did you even notice this? It looks like the error is in the info section which will be made better when POMDPs.jl 0.9 is released, so I wouldn't worry about it too much. As long as POMDPs is passing, I think it will probably be fixed in the future.
That being said, @lassepe you are probably right that we should register the solvers because they might be a pretty good test of the inference engine.
@zsunberg I was reading stuff in Base and saw this pop up. But this should actually not be an issue anymore as of JuliaLang/julia@73df43a. Thank you @JeffBezanson !