repo: avoid copying/updating a dict on every `repo.__getitem__` This has some mild performance benefits. I'm looking into a pathological case where one of our `hg log` invocations takes several seconds, and according to hyperfine this reduces the wall time of the entire operation (running in chg) from: ``` Time (mean ± σ): 7.390 s ± 0.106 s [User: 7.058 s, System: 0.271 s] Range (min … max): 7.300 s … 7.625 s ``` to: ``` Time (mean ± σ): 7.046 s ± 0.091 s [User: 6.714 s, System: 0.279 s] Range (min … max): 6.916 s … 7.169 s ``` Note: the log command is slow due to an issue in our custom stuff executing `repo[<arg>]` 298,800 times. This performance improvement is likely not noticeable during normal operation, but I don't feel like it's making the code more difficult to understand, and every small bit helps. Differential Revision: https://phab.mercurial-scm.org/D9022

obsstore: refactor v1 logic to fix 32 byte hash support Refactor the v1 logic to determine the node parsing based on the flag. Move the predecessor out of the fixed part and handle it like the other nodes, removing most of the duplicated code for parsing 20/32 bytes hashes. Differential Revision: https://phab.mercurial-scm.org/D8801

fsmonitor: increase the threshold before we recommend it, when using rust 50k files works just fine with the rust status, and it's annoying to get told "consider using fsmonitor" when we have another solution to the status speed. 400k files was not chosen in any rigorous way. I know 200k files is fine. Twice as many files should still be fine. Differential Revision: https://phab.mercurial-scm.org/D9021

copy-tracing: test case where a merge reverted a file deletion This case is currently broken for commit centric copy tracing. So we add an official case for it with a note about it being broken. Fixing it will requires multiples change in code related to merge, commit and copy tracing, so we introduce it beforehand for simplicity. Also, I have been sitting on this test for 3 months so I would rather have it upstream.