… for node ID → revision number lookups, instead on linear scan in a revlog.
Details
- Reviewers
Alphare pulkit - Group Reviewers
hg-reviewers - Commits
- rHG9eb07ab3f2d4: rhg: use persistent nodemap when available
Diff Detail
- Repository
- rHG Mercurial
- Lint
Automatic diff as part of commit; lint not applicable. - Unit
Automatic diff as part of commit; unit tests not applicable.
Event Timeline
rust/hg-core/src/revlog/nodemap_docket.rs | ||
---|---|---|
21 | I would add a comment explaining why it's ok for the docket to not exist, even if it's kind of obvious. | |
39 | 32 bit systems will have other issues before running into this one, I don't think it's worth it. | |
51 | I think we probably want to be able to disable mmap even in Rust contexts, although that will be done in a followup, since we need to have access to the config (storage.revlog.nodemap.mmap at the time of writing). | |
54 | I think the Python code has a bug, and that the idea is to silently ignore the missing file. | |
rust/hg-core/src/revlog/revlog.rs | ||
47 | Please add a docstring | |
116–117 | I think this comment should be adjusted now that the optimization has come | |
tests/test-rhg.t | ||
201 | I'd add a small header line to say "test partial node prefix" or something similar. |
Note to other reviewers, we're discussing moving at least some of the binary parsing in hg-core to a crate like zerocopy or bytemuck in the near future.
I would add a comment explaining why it's ok for the docket to not exist, even if it's kind of obvious.