Releases: Byron/trash-rs
v3.0.1
Chore
- bump
windows
crate to 0.44
Merge branch 'bump-windows-0.44'
Commit Statistics
- 4 commits contributed to the release over the course of 61 calendar days.
- 64 days passed between releases.
- 1 commit was understood as conventional.
- 0 issues like '(#ID)' were seen in commit messages
Thanks Clippy
Clippy helped 1 time to make code idiomatic.
Commit Details
v3.0.0
Chore (BREAKING)
- Upgrade from
windows
v0.37 to v0.43.
Commit Statistics
- 4 commits contributed to the release over the course of 132 calendar days.
- 145 days passed between releases.
- 1 commit was understood as conventional.
- 0 issues like '(#ID)' were seen in commit messages
Commit Details
v2.1.5
Bug Fixes
- Make chrono a default-enabled optional feature.
This allows to turn chrono support off without actually affecting the
ability to trash and restore items.
chrono
still has issues to dubious local-time support which relies
on a c-library function that can cause undefined behaviour as it
accesses an environment variable in a non-threadsafe fashion.
Commit Statistics
- 4 commits contributed to the release.
- 40 days passed between releases.
- 1 commit where understood as conventional.
- 1 unique issue was worked on: #39
Commit Details
v2.1.4
Fixes
- upgrade the
windows
crate to v0.37 to resolve a build issue and lay the foundation
for more regular updates of the windows support.
Commit Statistics
- 2 commits contributed to the release.
- 8 days passed between releases.
- 0 commits where understood as conventional.
- 2 unique issues were worked on: #39, #51
Commit Details
v2.1.3
Fixes
- include
windows
crate only on windows for reduced CI build times from ~9s to ~4s.
Commit Statistics
- 3 commits contributed to the release.
- 3 days passed between releases.
- 0 commits where understood as conventional.
- 1 unique issue was worked on: #5050505050
Commit Details
v2.1.2
Bug Fixes
-
avoid inconsistency when using relative paths in trashed file info.
We use absolute paths now without trying to generate a relative path
based on some top directory as the latter seems to be causing
inconsistencies on some linux distros, as the restore path ends
up being incorrect.Rather go with the absolute truth and don't fiddle with path
transformations at all to make it work everywhere.
Commit Statistics
- 1 commit contributed to the release.
- 2 days passed between releases.
- 1 commit where understood as conventional.
- 1 unique issue was worked on: #39
Commit Details
v2.1.1
Bug Fixes
-
Properly reconstruct paths when restoring files on freedesktop if those were relative.
Previously it would be unable to reconstruct original paths if the trash
directory was on a mount point due to a 'split brain' of sorts.When trashing files it would create original path information based
on them being relative to a mount point, but when restoring them
it would reconstruct them to be relative to the trash top level
directory.Now the reconstruction happens against to mount point itself which makes
restoration match.
Commit Statistics
- 6 commits contributed to the release over the course of 2 calendar days.
- 3 days passed between releases.
- 1 commit where understood as conventional.
- 1 unique issue was worked on: #47
Commit Details
view details
- #47
- Uncategorized
v2.1.0
Fixes
- Leading directories are now created on linux to avoid errors when trashing nested directories.
Commit Statistics
- 3 commits contributed to the release.
- 103 days passed between releases.
- 0 commits where understood as conventional.
- 1 unique issue was worked on: #45
Commit Details
v2.0.4
We detected the possibility of UB in the Linux and FreeBSD versions of get_mount_points()
and reduced the likelihood
of it happening in a multi-threaded environment by synchronizing access. You can read more about the state of
a more permanent fix in the tracking issue.
All previous 2.0.* releases which contained this function were yanked from crates-io.
Fixes
- Make internal
get_mount_points()
thread-safe to reduce chance of UB greatly.
This may reduce performance of crates that are using trash from multiple threads somewhat, as a part of the operation
is now synchronized. - Fix build on FreeBSD, handle UB similarly to the above.
Commit Statistics
- 9 commits contributed to the release over the course of 30 calendar days.
- 30 days passed between releases.
- 0 commits where understood as conventional.
- 0 issues like '(#ID)' where seen in commit messages
Commit Details
view details
- Uncategorized
- Prepare changelog for next release (b65f574)
- Add Mutex to linux version of get_mount_points(); document UB chance in lib.rs (c5c9c5e)
- Use Mutex to prevent concurrent access to getmntinfo (5c8e0ce)
- Merge pull request #43 from wezm/num-threads-freebsd (8f10c85)
- Fix build on FreeBSD after refactor (f3d31e5)
- Use
num_threads()
to avoid UB in FreeBSD version of get_mount_points() (3c153ae) - refactor (92ab7b9)
- Add BSD compatible implementation of get_mount_points (82d2132)
- Run
cargo-diet
for a more minimal crates package (561f21d)
v2.0.3
Bug Fixes
- let dependency specification in Cargo.toml match cfg directives in code
This fixes issue 40.
Commit Statistics
- 5 commits contributed to the release over the course of 125 calendar days.
- 1 commit where understood as conventional.
- 2 unique issues were worked on: #37, #40