On Wed, Oct 09, 2024 at 04:30:52PM -0400, Taylor Blau wrote: > The first optimization (cross-pack deltas) should help clones and > fetches, but the second optimization (thin deltas) will only help > fetches, since the 'haves' bitmap will necessarily be empty for > clones. > > Of course, REF_DELTAs have a less compact representation than > OFS_DELTAs, so the resulting packs will trade off some CPU time for a > slightly larger pack. But we're only talking about a handful of extra > bytes, and network bandwidth is pretty cheap, so I think the > optimization is worthwhile here too. It would be nice to see some numbers, even simulated ones from t/perf. Of course we'd like to show off any reduced server CPU for generating a fetch response. But I'd also like to see if the extra steps to find the cross-pack bases have any measurable negative effect (so the ideal there would be a big midx repo that doesn't have a lot of those bases, as it would not be helped much by the optimization and would be hurt by the time spent trying to apply it). -Peff
