Hello everyone! Over the last week, I started implementing generation number v2. Based on the performance numbers and subsequent discussion, we have decided to use generation data chunk approach, storing topological levels into CDAT and corrected commit date offsets into GDAT chunk. Do note that we are maintaining backward compatibility using topological level, not with monotonic offsets. I also found a performance regression with `commit-graph write --reachable --changed_path` that sneaked in along with the patch on moving generation and graph positions into commit-slab. https://abhishekkumar2718.github.io/gsoc/2020/07/12/gsoc-week-6.html https://github.com/gitgitgadget/git/pull/676 > Yes, I pushed without signing off or a cover letter. It's still a work > in progress. Thanks Abhishek
