summaryrefslogtreecommitdiff
path: root/crates/slotmap-careful
Commit message (Collapse)AuthorAgeFilesLines
* Bump minor versions of crates with new APIs/implsIan Jackson2024-10-301-1/+1
| | | | | cargo set-version --bump=patch -p slotmap-careful cargo set-version --bump=patch -p safelog
* Convert a few crates to slotmap-carefulNeel Chauhan2024-10-171-2/+14
|
* Minor bumps on non-tor/arti cratesNick Mathewson2024-09-301-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | These crates are not in the tor/arti namespace, but we have given them MSRV bumps: ``` oneshot-fused-workaround slotmap-careful fslock-guard hashx equix caret fs-mistrust safelog test-temp-dir retry-error ``` We are counting this as a breaking change. Since all of these crates are at 0.x.x, we have indicated the breaking change with a minor version bump. This commit was generated with the following script: ``` BUMPS=" oneshot-fused-workaround slotmap-careful fslock-guard hashx equix caret fs-mistrust safelog test-temp-dir retry-error " for crate in $BUMPS; do cargo set-version --bump minor -p $crate; done ```
* Upgrade MSRV to 1.77Nick Mathewson2024-09-251-1/+1
| | | | This will allow us to upgrade to the latest version of rusqlite.
* Bump MSRV from 1.70 to 1.75.Wesley Aptekar-Cassels2024-09-161-1/+1
|
* Fix typosDimitris Apostolou2024-09-031-1/+1
|
* slotmap-careful: Try to make test work with MSRV.Nick Mathewson2024-08-081-8/+4
|
* New `slotmap-careful` crate to use when we mustn't re-use keys.Nick Mathewson2024-08-084-0/+1300
This crate works as a drop-in replacement for the generational arena types `slotmap::{SlotMap, DenseSlotMap, HopSlotMap}`, and is implemented a set of wrappers around those types. The wrappers guarantee that slot versions numbers can never wrap around by marking as unusable any slot whose version number would otherwise get too high. (We add some leeway between our max allowed version number and the largest possible version number, so that we can detect bugs.) The code relies on the serde encoding of slotmap key versions. For notes on stability and (surprisingly good) performance, see the comments. Test coverage is around 98% for the lib.rs file; it's lower in key_data.rs, since the error cases are unreachable given slotmap's current behavior. Open questions: * What further testing is a good idea? * Will slotmap ever upstream something like this? See "# Limitations" comment for the parts of slotmap that are not implemented; I hope that we don't need them.