We only benchmark the latest known commit when a git repository is updated... but the git push could have added more than one commit at the same time. If we don't benchmark the intermediate commits, then the graph timeline is harder to read as it doesn't reflect the true git history.
As a bonus, this might help with retro-benchmarking: We currently have to wait for each benchmark to complete, rather than submitting multiple commits corresponding to the different versions in one go!
We only benchmark the latest known commit when a git repository is updated... but the git push could have added more than one commit at the same time. If we don't benchmark the intermediate commits, then the graph timeline is harder to read as it doesn't reflect the true git history.
As a bonus, this might help with retro-benchmarking: We currently have to wait for each benchmark to complete, rather than submitting multiple commits corresponding to the different versions in one go!
(related to the removal of old commits https://github.com/ocurrent/current-bench/issues/378 which looks a bit easier to handle first)