@benmac006@mcgeedan42@ycombinator what exactly are optimizations? Are the results from cold runs? You made some heavy claims with very little info on how exactly you achieve any of it.
@benmac006@mcgeedan42@ycombinator How does one repro any of the benchmarks? For something like Go, what granularity do you consider a test? If not package, how do you deal with shared globals, state, test ordering or subprocess behavior. Cargo test also works like this iirc. Is the graph static only?
@aravindputrevu Code review is alot more likely lmao. the need for verification wont go anywhere. That’s all ci is. People just put terrible gates in it then complain lmao like only unit/integration testing
@rs545837@aravindputrevu the need for verification wont go anywhere. That’s all ci is. People just put terrible gates in it then complain lmao like only unit/integration testing
@colek42c@tonychang430 I see. Directionally agree, i think integrating this with bazel or buck could be interesting and maybe a few examples to make it easier for folks to adopt. I think this will need to be mainstream soon forsure
@colek42c@tonychang430 intriguing but this doesnt avoid needing ci or larger machines for anything bigger than your working mac? i wonder what the split of what can be run locally vs scaled out
@rs545837@tonychang430 does this use ebpf? How does this work for test runners that dont expose a way to isolate the specific tests so you can skip them?