NewYour coding agent can read the release notes before it upgrades.Set up the MCP server →
crates.io
crates.io, #2285 most downloaded on crates.io
A crate of the gitoxide project to provide common errors and error-handling utilities
Latest stable
0.5.2
Last release 1 months ago
22 Oct 2026
Ships fairly regularly
a new release about every 3 weeks
Nearly every release is documented
notes for 15 of 15 stable releases
Nothing withdrawn
no release was ever pulled
10 months old
15 releases · first in 2026
Drag across the ruler to compare, or tap a release to open it.
3 commits contributed to the release over the course of 13 calendar days.
Your coding agent can read these notes before it upgrades. Set up the MCP server →
Document the gix-utils/prodash workaround and guard auditable builds
Document the gix-utils/prodash workaround and guard auditable builds
cargo-auditable reconstructs enabled features from Cargo’s rustc arguments
and replays them into workspace-wide cargo metadata. The root package’s
implicit prodash feature also matches gix-utils’ optional dependency,
despite dep:prodash suppressing its implicit feature. Cargo then rejects the
request for the nonexistent gix-utils/prodash feature.
Explain why the explicit alias is a tooling compatibility workaround, link
the upstream Cargo and cargo-auditable issues, and document a metadata
reproduction and how to reassess removal.
Add just check-auditable and a blocking CI job using stable Rust and
cargo-auditable 0.7.6. Test an actual auditable build of the small
configuration rather than requiring the alias itself, so an upstream fix can
allow the workaround to be removed without rewriting the check.
gix-utils/prodash workaround and guard auditable builds (6a262bc)skip classification markers in flattened error chains
ClassificationMarker supplies recovery metadata without a diagnostic of its
own. Flattened error chains exposed these markers as extra causes: an
intrinsically classified leaf gained a trailing Validation or Conflict,
and a source-bearing tag repeated the diagnostic of the error it wrapped.
Skip markers during ChainedError::source() traversal and promote the first
real diagnostic when converting a marker-rooted Exn to anyhow::Error.
Align the public Error source chain in auto-chain-error mode with its
existing display promotion so tagged roots appear only once. Keep markers
stored internally for classification and tree recovery, and preserve the
root display fallback for marker-only errors.
avoid redundant Unicode precomposition work
avoid redundant Unicode precomposition work
We now preserve macOS conversion failures per filename, but splitting
every path adds work even for ASCII. In the initial benchmark, ASCII paths
took about 2.6-2.8 times as long as before. The existing is_nfc()
check can also perform a full NFC conversion before our compatibility
composer performs its own decomposition and composition.
Use is_nfc_quick() and detect non-BMP characters through the same iterator.
Only scan its unvisited suffix if quick-check stops early. ASCII strings
return immediately; ASCII paths also avoid UTF-8 validation and splitting.
Process BMP-only paths together, reserving component splitting for paths
with conversion failures. Specialize filename and path handling at compile
time to keep the path fallback out of the filename loop.
Skip combining-class and composition lookups for ASCII starters, use byte
lengths for buffer capacities instead of counting characters, and reserve
the output string before encoding it. Keep noncanonical mark order, Hangul
composition, component-local fallback, and the original Cow when unchanged.
The capacity estimate can reserve more space than the previous character
count. Inconclusive NFC checks can now use a temporary buffer even if the
result is unchanged; the benchmark explicitly covers that case. There are
no new runtime dependencies; Criterion is already used in this workspace.
Add a Criterion benchmark covering filenames and paths with ASCII, NFC,
NFD, Hangul, noncanonical marks, and mixed emoji/decomposed components.
| Case | Original HEAD median (ns) | This commit median (ns) |
|---|---|---|
filename/ascii |
6.46 | 2.55 |
filename/emoji_first |
63.54 | 3.17 |
filename/emoji_last |
57.95 | 15.80 |
filename/hangul_nfc |
12.20 | 11.88 |
filename/hangul_nfd |
141.87 | 70.28 |
filename/nfc |
16.60 | 15.78 |
filename/nfc_quick_check_maybe |
70.32 | 32.75 |
filename/nfc_with_mark |
11.03 | 11.89 |
filename/nfd |
303.23 | 70.96 |
filename/noncanonical_marks |
46.98 | 32.58 |
filename/several_nfd |
139.12 | 58.56 |
path/ascii |
34.71 | 2.69 |
path/ascii_long |
98.55 | 4.23 |
path/mixed |
403.60 | 134.15 |
path/nfc |
50.75 | 26.13 |
path/nfd |
378.76 | 234.88 |
path/reported |
106.27 | 37.88 |
path/several_nfd |
343.42 | 101.15 |
The NFC string with an uncomposable mark costs about 0.9 ns more; the
other measured cases improved. These are normalization microbenchmarks,
not measurements of complete status operations.
To compare subsequent changes:
cargo bench -p gix-utils --bench precompose -- --save-baseline before --sample-size 100 --warm-up-time 1 --measurement-time 2 --noplot
# Apply the change being measured, then:
cargo bench -p gix-utils --bench precompose -- --baseline before --sample-size 100 --warm-up-time 1 --measurement-time 2 --noplotpreserve filenames rejected by macOS precomposition
The original Starship report states that a tracked filename
containing both an emoji and a decomposed umlaut is reported
as untracked by gix status.
Git's macOS UTF-8-MAC conversion rejects characters outside the Basic
Multilingual Plane. When conversion fails, Git keeps the entire original
input, including any decomposed umlauts before or after the rejected
character. Our precomposition instead changed those umlauts, preventing
the directory walk from matching Git's index entries.
Preserve inputs containing non-BMP characters in the shared precomposition
helper. Apply that rule separately to each filesystem-path component,
preserving path separators and borrowing unchanged paths. Reference packing
and lookup also use component-wise precomposition, consistent with loose
reference iteration.
Explicit file roots in the directory walk use the same path helper, so
their emitted spelling agrees with ordinary traversal beneath emoji
directories. Extend the existing root-precomposition test to cover an
emoji parent and an emoji filename, with precomposition enabled and disabled.
An emoji outside a reference name must not suppress that name's
precomposition: otherwise loose and packed versions of an umlaut branch
can be emitted twice. Cover that review finding, packing and lookup below
an emoji component, both emoji/umlaut orderings, another supplementary-plane
character, the reported Teaching/🎥 Überwachung im digitalen Zeitalter.md
filename, and BMP-only controls.
Restore the requested reference spelling after lookup instead of
decomposing the entire returned name. Whole-name decomposition could alter
an untouched emoji/umlaut component and make the returned name unusable
for another lookup or deletion. Cover mixed composition, loose and packed
references, full names, and worktree-qualified lookup round trips.
Keep canonical pseudo-ref names when precomposition changes their
classification, including worktree-qualified pseudo-refs.
Both the helper regression and macOS status regression failed before the
fix and now pass, as do the reference regressions added for the review
finding. Validated the gix-utils, gix-fs, gix-dir, gix-status, and
gix-ref suites, the gix status tests with default features and with
--no-default-features --features basic,extras,comfort, targeted Clippy
with warnings denied, and cargo fmt --all -- --check.
Git reference: Apple Git 2.54.0 with core.precomposeUnicode=true leaves
the repository clean. /usr/bin/iconv -f UTF-8-MAC -t UTF-8 rejects these
non-BMP characters. compat/precompose_utf8.c::precompose_utf8_readdir()
and precompose_string_if_needed() at
d38352cd43ab9745686d697872408bc3249a153f retain the original input on
conversion failure.
Maintenance release without any new features.
<csr-read-only-do-not-edit/>
<csr-read-only-do-not-edit/>
<details><summary>view details</summary>
9cb09b2)5c7b52d)2571113)aacf0f0)ca5b801)
</details><csr-id-1f56236675ef84daa5b5acbab0a1555efec9f593/> make caller-location printing opt-in
<!-- agent -->
auto-chain-error also printed source locations in user-facing reports,
including consumers which only wanted a standard error chain. Make printing
independently configurable with error-print-location, forwarded by gix.
The feature gates the shared location writer; caller locations, metadata,
classification, and the existing alternate-format rules remain intact.
Enable printing explicitly in workspace binary dependencies and the filter
test-helper feature. gix-error opts in for its own tests so existing location
snapshots retain their coverage. An isolated Cargo consumer checks real
main() stderr for both crates, printing on and off, tree and chain modes,
and tree-error precedence without workspace test feature unification.
<csr-id-a347e617e7bd1304e9518a9a182e19194b833d47/> shorten caller locations to package source paths <!-- Byron -->
Reviewed and refactored carefully, tried the release script as well. Didn't review the release script at all though, just had some docs added. I consider it disposable.
<!-- agent -->
Enabled location diagnostics could expose entire checkout and Cargo registry
paths, obscuring the useful source location with machine-specific directories.
Format conventional source paths relative to their containing package directory
and remove Cargo registry version suffixes, so a path such as
/home/user/.cargo/registry/src/index.crates.io-hash/gix-url-0.39.0/src/parse.rs
prints as gix-url/src/parse.rs. Unrecognized layouts fall back to the filename.
Keep captured std::panic::Location values and line numbers intact. The shared
formatter covers tree, chain, source, test, and main() diagnostics; feature
defaults and alternate formatting remain unchanged. Document the limitation
that locations contain no Cargo package metadata and build owners can use
--remap-path-prefix for exact package names in arbitrary layouts.
<csr-id-d18a82b1cafa15d01d02cabefff1cf580f47902b/> preserve explicit causes in standard error source chains
<!-- agent -->
Returning a child frame's bare diagnostic from Error::source() lost any
explicit causes attached to that frame. Standard error reporters, including
anyhow, could truncate nested context chains before reaching the underlying
error.
Keep nonleaf frames reachable as source boundaries while exposing raw leaf diagnostics for downcasting. Format each boundary as a single diagnostic to avoid duplicating subtrees in external reports, and preserve native-source precedence without another allocation.
Add regression coverage for standard and anyhow traversal and rendering,
erasure and boxing round trips, and preservation of typed diagnostics,
classification, and metadata. Document linear source traversal and the
full-tree inspection alternatives.
<csr-id-23321850e43c434cebd4eeaba40462af0033c178/> flatten native error chains in linear time
<!-- agent -->
Exn::into_chain() resolved each native source from its owning root both
to check for nested boundaries and to construct the next handle. Flattening
64 errors therefore required 4,096 source() calls.
Scan each native chain once when constructing its ErrorHandle, recording
its length and the prefix that inherits the frame location. Advancing a
handle only increments its depth. Stop at nested Error boundaries to
preserve breadth-first order, concrete error types, and caller locations
without extra allocations or dependencies.
Add native_source_flattening_is_linear to require at most four source
lookups per error and separately verify that every error survives.
Chains of 16, 32, and 64 errors now need 16, 32, and 64 source calls.
Explicit into_chain() calls exercise flattening in every feature
configuration. Individual handle lookups continue to walk from their owner.
<csr-id-aa7b69919317fdef6882895deb03988699c72859/> wrap Metadata and simplify diagnostic keys
<!-- agent -->
Quoting every metadata key adds noise to diagnostic output even when the
name is already unambiguous. Keep nonempty keys containing only ASCII
letters, digits, _, -, or . unquoted in Message display and metadata
debug output. Empty and unusual keys remain quoted and escaped, and value
formatting is unchanged.
Replace the public BTreeMap alias with a private-storage Metadata
wrapper so dictionary formatting and its public API no longer expose the
implementation. Preserve lexicographic ordering and provide construction,
lookup, mutation, iteration, indexing, collection, and extension APIs.
Breaking change: Metadata is no longer interchangeable with
BTreeMap. Callers must use the wrapper API; iterator return types hide
the underlying collection.
Add regression coverage for key quoting, compact values, and dictionary access, and update dependent assertions and output snapshots.
<csr-id-506fc1df9ad7931139503a80d3a51aa0117a0ef8/> unify workspace diagnostics and error recovery <!-- Byron -->
Looked at gix-error carefully and did multiple cleanup and improvement passes.
At some point I went quicker through the changes as they looked good enough,
also to at some point just wave it through.
<!-- agent -->
Use the workspace's own error API throughout applications, examples, internal
utilities, and fuzz targets. Replace downstream anyhow::Result, Context,
bail!, and ensure! usage with gix::Result and helpers from gix::error,
or gix_error in plumbing crates. Convert native errors at their call sites
while preserving concrete causes, typed recovery contracts, and caller locations.
Keep optional anyhow interoperability in gix-error and gix for consumers.
Make diagnostics useful both to people and to callers deciding how to recover. Distinguish invalid input, corrupt data, missing resources, unavailable capabilities, authentication, authorization, conflicting state, and cancellation. Explicit cancellation prevents retrying even when another cause is retryable. Store offending input and available subprocess details as structured metadata, with operation context in the message and native failures retained as causes.
Report causes under a numbered Caused by: section, retaining branch guides for
error trees and flattening linear chains. Standardize path quoting, refresh
snapshots, and omit changing source locations from CLI error comparisons.
Migrate affected tests and helpers to TestResult where they need to propagate
both native errors and exceptions. Update contributor and application guidance,
remove redundant dependencies, and refresh the lockfile.
gix-error: recovery classes, metadata, and complete native reportsAdd Cancelled, PermissionDenied, Unauthenticated, Conflict, and
Unsupported classes with matching constructors, markers, builders, and
predicates. Order classes by suggested recovery precedence and expose
dominant_class(). Both retry policies inspect cancellation across causes;
native permission and unsupported I/O errors receive classifications, while
explicit authentication challenges take precedence over legacy permission wrappers.
Add classified Message builders and _error helpers, caller-tracked
Message conversions, and formatted bail!/ensure! shorthand with builder
chains. Introduce with_input(), subprocess metadata helpers, common metadata
schemas, and metadata_merged(), with more specific causes overriding context.
Number tree reports while preserving forks and flattening chains. Give
auto-chain-error complete Debug reports with locations and single-line
alternate Display chains; share that formatter with TestError. Cover nested
native sources, classification precedence, metadata, caller tracking, macro
evaluation, report layout, and retained anyhow interoperability.
gix: consistent porcelain diagnostics and application guidanceDocument gix::Result, gix::Error, and gix::error for applications. Apply
classification and input metadata to configuration, initialization, repository
opening, clone and remote setup, object access, and path handling. Distinguish
unsupported formats from malformed values, and classify branch/worktree state
conflicts and explicit cancellation without replacing concrete recovery errors.
Record signing and key-command failures through subprocess metadata. Migrate
examples and tests away from anyhow, refresh diagnostics and classification
assertions, remove the test dependency, and update prodash to 31.0.1.
gitoxide: native CLI errors and dependency cleanupReturn gix::Result from gix, ein, the combined binary, and shared command,
progress, and tracing helpers. Preserve context when configuring repositories,
writing output, and starting tracing; classify unavailable tracing as unsupported.
Migrate top-level examples and journey snapshots to native diagnostics.
Replace is-terminal with std::io::IsTerminal, remove anyhow, and update
prodash to 31.0.1 with corresponding feature and lockfile cleanup.
<csr-id-7cbd9eb87e822db146c09c4403851330886d71ef/> return public errors from extension helpers <!-- Byron -->
gix-error was refackiewed, the rest was scrolled through to peek out problems
at best.
Looked at gix in detail, refackiew, and came up with API improvement to
gix-config-value on the way.
<!-- agent -->
Public plumbing APIs now use gix_error::Result, so constructing typed
exceptions by default requires repeated erasure, conversion, and propagation
wrappers at their call sites.
Make raise() and and_raise() return Error, and make or_raise() and
ok_or_raise() return Result. Preserve the former behavior under explicit
_typed names and add ResultExt::or_error() for conversion without context.
Existing public errors pass through unchanged without allocating, while
context retains concrete causes, metadata, classifications, and caller sites.
Keep the erased helpers and inherent Exn operations unchanged.
Migrate callers and documentation together, preserving concrete public
exception signatures and typed internal boundaries. Use the existing bail!
macro for 388 early error returns across 37 crates, passing native errors and
messages directly and retaining context and explicit erasure on typed
exceptions. Preserve direct propagation of public Error values to avoid
introducing extra exception frames.
<csr-id-ac71fb42cd2417d27dfd0c9f4ec769abd1e95839/> standardize public errors and reuse exception frames <!-- Byron --> rubber-stamp after scrolling through the diff top to bottom. gix-error changes were refackiewed carefully.
<!-- agent -->
Use gix_error::Result and gix_error::Error at public plumbing boundaries
with erased or message-based exceptions so applications can use ordinary
std::error::Error handling while retaining concrete recovery errors,
causes, metadata, and caller locations. Apply this to public traits,
callbacks, iterator items, and re-exported APIs, and adapt their consumers
throughout the workspace. Preserve concrete public error signatures and
typed internal exceptions; let bail! convert to either result form.
Add Error::into_exn() for internal tree manipulation, including the
revision parser's delayed errors. Recover explicitly raised frames in
auto-chain-error mode and forward native source chains through nested
Error wrappers. Reuse existing boxed tree frames when adding context or
erasing errors to avoid redundant boxing, and keep context construction
independent of existing chain depth.
Remove repeated propagation wrappers from callers and update the migration guide and repository conventions. Document the value of context that explains an operation's intent or identifies user-controlled configuration, including its raise location, even when failures are rare.
Remove Class::Io and the synthetic io() constructor so classifications
describe semantic failures. Retain NotFound and OutOfMemory
classifications for real I/O errors, and inspect their concrete causes for
the conservative and lenient retry policies.
<csr-id-a36bff1d40022d35820cbea693bbad093e631464/> replace recovery tags with typed operation errors <!-- agent --> GitButler needs to distinguish missing binary merge results, ambiguous object prefixes, diff setup failures, unrelated histories, and repository discovery outcomes. Broad classifications alone also match failures that these callers must propagate, such as missing-directory I/O and rejected repository trust.
Remove Class::Tagged and expose non-exhaustive Error enums for these
operations. Share the merge-base enum across pair, many, and octopus methods,
and keep porcelain returns as gix::Result. Preserve candidate information,
native paths, diff resource details, requested commit IDs, and callee causes.
Add gix_error::tag() and make classifications identify the wrapped error or
the owner of a constant marker source. Keep traversal lazy and preserve its
order, duplicates, marker transparency, and original I/O kinds in both error
representations. Document typed recovery and wildcard propagation.
Validate merge-base input commits in the shared traversal path so missing
inputs cannot masquerade as unrelated histories, including fast paths and
octopus calls. Continue allowing absent parents for shallow histories and
empty successful results for merge_bases_many*() on unrelated histories.
Fix the recovery tests failing in Linux/ARM test-fast and Windows
test-fixtures-windows CI. Discovery examines the ceiling directory before
stopping, so its immediate child reports a stopping height of 2. Attribute
lookup converts Git paths to native paths, rejecting ill-formed UTF-8 on
Windows before reaching object lookup. Cover missing-object conversion with
a valid path everywhere and retain its non-UTF-8 path case on Unix.
<csr-id-f0940c590e7142820ac191198f4539b31e48bc42/> make bstr an opt-in metadata conversion feature
<!-- agent -->
Small error-handling programs should not compile bstr and memchr just
to add ordinary error contexts. Disable bstr by default so a standalone
gix-error consumer needs no dependencies unless it enables optional features.
Always store MetadataValue::Bytes as Vec<u8>, regardless of features.
This changes the public variant payload from BString. Keep escaped debug
and display diagnostics compatible with bstr, including valid Unicode,
control characters, invalid bytes, and truncated UTF-8.
The opt-in bstr feature supplies From<BString> and From<&BStr> for
MetadataValue and retains the crate re-export. Owned conversions reuse
the allocation; borrowed conversions copy the bytes. Enable the feature
explicitly in workspace crates that need these conversions or the re-export,
preserving their with_input() and with() calls. Adapt direct byte
constructors and submodule payload access to the new Vec<u8> representation.
Earlier five-sample fresh-build measurements of the same small consumer on
Rust 1.99.0 / M4 Max recorded medians of 702 vs 458 ms in debug and 872 vs
609 ms in release with and without bstr. These include Cargo and linking
with registry sources cached; they are specific to that consumer and machine.
<csr-read-only-do-not-edit/>
<csr-read-only-do-not-edit/>
<details><summary>view details</summary>
0115e90)794eeef)1d7bac7)a347e61)1f56236)d18a82b)Metadata and simplify diagnostic keys (aa7b699)9272d45)bstr an opt-in metadata conversion feature (f0940c5)80f4b03)506fc1d)f819565)2332185)7cbd9eb)ac71fb4)a36bff1)5fb3dcf)
</details>Clone whole packs with gixp --verbose pack-receive https://github.com/git/git --refs-directory out/all-refs out ![asciicast]
gixp --verbose pack-receive https://github.com/git/git --refs-directory out/all-refs out
gixp --verbose remote-ref-list https://github.com/torvalds/linuxFor some reason these are currently broken and I don't want to spend time on fixing it right now. PRs to fix this are very welcome.
git_is_space()raise MSRV to Rust 1.88
The newly published dua-core 3.3 release used by linked-worktree removal
requires Rust 1.88, so raise every workspace crate and the advertised badge
together.
Keep the MSRV checks buildable by selecting the latest sysinfo and rusqlite
release lines that support Rust 1.88.
move the sirocco feature to sirocco 0.2
sirocco 0.2 decodes into a buffer the caller owns, in one call or many,
so it now handles all zlib decompression with the feature, with no
fallback to zlib-rs:
Inflate::once() and Inflate::prefix() decode with sirocco. Asirocco's partial output.Inflate::rest() continues the stream prefix() started, whichInflate::state.zlib-rs as before.stream::inflate::Measure uses sirocco::inflate::Scanner.sirocco's errors map to the classes and messages zlib-rs produces,
corruption as "Invalid input data" and a preset dictionary as a
validation error.
add the sirocco feature to decompress with sirocco
sirocco is a pure-Rust DEFLATE/zlib decompressor that is faster than zlib-rs
on whole buffers, and can find where a stream ends without producing its output.
With the feature enabled:
Inflate::once() decodes whole streams with sirocco, and falls back to zlib-rsInflate::prefix() always uses zlib-rs, for callers that only want thegix-pack and the loose object header.stream::inflate::Measure finds the compressed and decompressed size of aBufRead. BytesToEntriesIter uses it instead of decompressingio::sink(). With sirocco, the decompressed bytes aren't produced, so thezlib-rs is still used for everything else, including all compression.
The gix crate exposes it as zlib-sirocco, and the fast feature of gitoxide
enables it for the gix and ein binaries. sirocco requires Rust 1.89.
hide the Inflate backend and align stream lifecycles
Spend a lot of time with this one and massaged it into place, overall simplifying
a little and adding test-coverage (that I am sure is already in Sirocco but doppelt
haelt besser).
The public Inflate::state field exposed a zlib-rs decoder even when
sirocco handled all decoding. Make backend state private and retain only
the selected implementation, avoiding an unused zlib-rs allocation when
sirocco is enabled.
Inflate backend and align stream lifecycles (0ed7c8c)sirocco feature to sirocco 0.2 (2b4faf9)sirocco feature to decompress with sirocco (95a0b14)remote-ref-list and pack-receive subcommands to gixindex-from-pack to pack-index-from-data<csr-read-only-do-not-edit/>
<csr-read-only-do-not-edit/>
<details><summary>view details</summary>
fb1f048)445be27)46a3511)c06d819)487d06d)2fcb557)b38290e)35e26fc)f90b92f)161e7df)5ea49c8)40a6412)386673c)b4a6e16)
</details><csr-id-dcdddd9429e3faf2083be340f9ec2bfaab232c2f/> attach scalar metadata to error contexts
<!-- agent -->
Callers can enrich errors with named values they already possess without
introducing a custom payload type. Metadata keeps a message and an ordered
dictionary of typed scalar values, including lossless bytes and native paths.
<csr-id-82776d93ab7671729b80ad89e420236c93cd1651/> inspect borrowed error classifications lazily Admittedly, I was mostly checking the public API, thinking that it's most certainly useful and overally, having lazy iterators for everything is a step in the right direction.
This wasn't a crazily detailed review in the interest of time.
<!-- agent --> Error inspection rebuilt the entire error graph before returning its first item, so even a root match visited unrelated sources and allocated storage. Use a shared iterator that expands each node only when another item is needed, preserving breadth-first order, concrete types, and caller locations in both error representations.
Expose classify(&error) for custom borrowed errors, and share classification
predicates with Error and Exn. The shared traversal and predicate definitions
remove more code than the borrowed API adds. Keep probable_cause() unchanged.
<csr-id-4fa58c5aee832a70372ad6c02f8573b8284bf040/> expose borrowed exception retry policies
<!-- agent -->
Add Exn::can_retry() and Exn::can_retry_lenient() using the existing
classification and retry rules. Callers can now inspect retry policies
without consuming an exception to produce Error.
Cover native sources, nested errors, explicit markers, all recognized I/O retry kinds, and allocation failures in both error storage modes.
<csr-id-ef8ced5db1c668efaff7485e501f78be447b88d0/> expose borrowed exception classifications
<!-- agent -->
Add Exn::classify() so callers can inspect normalized classes and their
original errors without converting the exception into Error. Reuse the
existing collection and classification logic, preserving breadth-first
ordering through nested boundaries, duplicate classes, and I/O origins.
Route the exception predicates through that shared classification path.
<csr-id-ce38e3b6529b4df52631b02c6116580ddf4d73cd/> simplify borrowed error inspection
<!-- agent -->
Expose semantic predicates on Exn and Error so callers can classify
failures without downcasts or ownership-consuming conversions. Preserve
explicit retry markers as distinct from the I/O retry policies.
Make nested error lookup follow logical breadth-first order and add
borrowed Exn::probable_cause(). Distinguish native-source identities even
when a source shares its owner's address, preserving the selected cause
through conversion. Reuse the existing inspection collectors.
Consolidate representation-independent helpers and document conversions that retain context and caller locations.
<csr-id-ac88bd7307b584183815e124b86fa44ac59b3988/> preserve classifications in custom I/O payloads
<!-- agent -->
std::io::Error::source() skips its custom payload, hiding classification
markers, custom error types, and branches of a nested gix_error::Error.
Retain the payload while walking native sources so borrowed retry policies,
exception traversal, and both porcelain error modes see the complete cause.
Document how custom errors expose their immediate cause and classification markers. Cover every classification, nested branches, and concrete payload downcasts, and update diagnostic snapshots for the retained I/O payload.
<csr-id-ca0f05dbefd75977f72485533e140015295f9590/> render nested error causes once
<!-- agent -->
Rendering a nested Error invoked its complete formatter before expanding
its causes again, duplicating messages and mixing formatting modes. Render
the concrete root once and skip that exact root during nested expansion;
its position is not necessarily first after filtering nested boundaries.
Apply the selected formatting mode and caller locations to nested causes.
<csr-id-c28e49464db05f2b510db13dfc60384846b36cfb/> keep drained exceptions safe to inspect
<!-- agent -->
Converting a typed Frame into a bare Exn left its error box typed while
error(), dereferencing, and extraction expected Untyped, causing a
panic after drain_children(). Wrap typed roots during the shared frame
conversion while retaining existing erasure, sources, and child frames.
<csr-id-da0f21d3f48266eb47a23548d5ae4163b78651dc/> unify diagnostics as Message and preserve causes
<!-- Byron -->
I refackiewed gix-error with the usual care, and skimmed through all the downstream changes, rubber-stamping it. Definitely cleaned up a few bits, like config-key related error handling, and error related types.
<!-- agent -->
A single failure could require separate errors for its message, category, and diagnostic values. Classification wrappers could obscure the concrete cause, while cause-selection heuristics could choose a different sibling after adding context or flattening an exception tree. Represent each failure with one diagnostic and preserve its actual causal structure.
Make Message the shared diagnostic type, with a message, optional Class,
and named MetadataValues. Classified constructors and the with_class()
and with() builders attach categories and details without extra error
layers. Plain messages and string conversions remain unclassified.
Metadata is now the value dictionary; Exn::metadata() and
Error::metadata() yield only non-empty dictionaries in traversal order,
keeping each context separate and preserving bytes, paths, and numeric types.
Introduce ClassificationMarker to classify existing concrete errors or
provide classification-only sources. Classification still inspects these
markers, while diagnostic traversal, downcasts, cause selection, and
reports skip them and retain their real descendants. Preserve concrete
I/O errors and their original kinds for recovery decisions.
Give Frame, Exn, and Error the same probable_cause() policy: follow
the unique causal path to a leaf or the first diagnostic branch, retaining
the aggregate instead of choosing an arbitrary sibling. Include native
sources, I/O payloads, nested errors, and explicitly raised children in both
tree and chain representations. Preserve caller locations through transparent
markers, avoid duplicate native causes in reports, and honor alternate
TestError formatting. Marker-only errors retain a classification fallback.
Add ExnResult<T = (), E = exn::Untyped> and ExnMessageResult<T = ()>,
re-export them from gix, and adopt them throughout the workspace. Keep
callback bounds erased and preserve concrete error types where callers
need their payloads. Group supporting classification and display types in
gix_error::types, and frame inspection and erasure in gix_error::exn.
Migrate constructors, signatures, and recovery checks together. Replace the
gix::config::key::Error family with ordinary diagnostics and central
gix::Error results. Provide config::key::error() and error_with_value()
for key, optional environment_override, and optional input metadata.
Retain parser causes, rejected numeric values, accepted special values,
and configuration leniency. Collapse artificial error layers elsewhere
while preserving concrete failures and partial outcomes.
Make complete diagnostics reviewable through inline insta snapshots,
retaining assertions for classifications, retry policies, concrete causes,
and stored data. Use the shared snapshot redaction helper for unstable
paths, ports, object IDs, and platform-specific OS errors. Update migration
guidance, examples, and CLI snapshots, and use static X.509 assertion messages to
avoid dumping verifier identities and raw output.
<csr-id-4b42e0ce80ae934cae4f102f44c392581758608f/> raise MSRV to Rust 1.88
<!-- agent -->
The newly published dua-core 3.3 release used by linked-worktree removal
requires Rust 1.88, so raise every workspace crate and the advertised badge
together.
Keep the MSRV checks buildable by selecting the latest sysinfo and rusqlite
release lines that support Rust 1.88.
<csr-id-b7ebd86cea6043224316b15abf90890e9214a5f0/> expose structured error classifications
Marked breaking because can_retry() now has a much more conservative default.
So downstream has to adjust and use can_retry_lenient() if this is desired.
<!-- agent --> Add ordered, non-exclusive classifications that retain the concrete error source. Resource exhaustion now distinguishes configured allocation limits from allocation or capacity failures, while existing boolean helpers remain compatibility views.
<csr-read-only-do-not-edit/>
<csr-read-only-do-not-edit/>
<details><summary>view details</summary>
37860b3)6356013)Message and preserve causes (da0f21d)dcdddd9)82776d9)ac88bd7)4b9ff51)4fa58c5)ef8ced5)ca0f05d)c28e494)ce38e3b)a095334)b7ebd86)4b42e0c)7e35849)
</details>add ergonomic test errors and string comparisons
<csr-read-only-do-not-edit/>
<csr-read-only-do-not-edit/>
<details><summary>view details</summary>
888677a)e3a6fa1)870c191)6704303)
</details>add BoxedResultExt for boxed errors
<csr-read-only-do-not-edit/>
<csr-read-only-do-not-edit/>
<details><summary>view details</summary>
5 commits contributed to the release over the course of 13 calendar days.
pack-explode and pack-index-from-data sub-commandspack-verifyMany small and possibly breaking changes are not mentioned here.
<csr-read-only-do-not-edit/>
<csr-read-only-do-not-edit/>
<details><summary>view details</summary>
34b2373)e71da0f)f59f66e)763d7ca)3a4faec)1931575)181d69c)e509373)d7591e2)c0352c2)91ba045)bca7ce2)584350a)395c7e7)8812e91)657aa2c)1e2ec7e)cc147bc)43c7fd1)9660bbf)44b8221)665696f)ee45c7f)0a33b24)29b9c23)adbc82c)0d31ad1)1805d64)0bcb790)f66b116)f932256)ec8c48a)d3c00c8)5e23137)93a9b30)aeb8778)bb45c3d)ec40e09)bcb275e)41f38c4)a7cfac8)afae684)fbb9f98)2888f1b)f98c5b1)7c5d8b8)
</details><csr-id-1af3e724c305862975bab35a6c9c263abb5d894b/> preserve and classify typed error sources <!-- agent --> Error classification and probable-cause selection need to inspect the complete error graph without turning native sources into strings or behaving differently across tree and auto-chain modes.
Add CorruptionError, NotFoundError, and RetryableError together with Error classification helpers and support for boxed standard errors. Classification follows native source chains and nested gix errors while retaining concrete types for downcasting.
Keep native sources owned by their original errors and traverse them lazily alongside explicit frames. Add breadth-first error iteration with optional captured locations, direct stored-error access, and downcasting across the complete graph.
Preserve probable-cause identity and logical parent relationships while flattening exception trees into ChainedError. This lets auto-chain mode reconstruct the same traversal order without duplicating nested compatibility source chains.
<csr-read-only-do-not-edit/>
<csr-read-only-do-not-edit/>
<details><summary>view details</summary>
keep erased errors visible to source-iteration and downcasting Since 499402c941, erasing an Exn wraps its error in the Untyped marker so that the type
<csr-id-ccd5bd412eeaf3fa9537a021d647f818d4a27833/> keep erased errors visible to source-iteration and downcasting Since 499402c941, erasing an Exn wraps its error in the Untyped marker so that the typed accessors of Exn<Untyped> keep working. However, the marker also hid the original error from everything that walks the error afterwards: frames now yielded the marker, which cannot be downcast to the original type, and whose empty Error implementation cut off the source chain below it.
This broke gix diff file, whose fallback for treating a revspec as a path on disk downcasts the sources() of a failed rev-parse to find the ref-not-found error - it would now fail with "couldn't parse revision" instead.
<csr-read-only-do-not-edit/>
<csr-read-only-do-not-edit/>
<details><summary>view details</summary>
5 commits contributed to the release over the course of 28 calendar days.
<csr-read-only-do-not-edit/>
<csr-read-only-do-not-edit/>
<details><summary>view details</summary>
10c58bb)ab2fee1)2cb17b2)3675a8d)adb8328)
</details>improve error message around "Signature name or email must not contain..." This might make it easier to understand where the error is coming from.
<csr-read-only-do-not-edit/>
<csr-read-only-do-not-edit/>
<details><summary>view details</summary>
76a03eb)53f880c)4d5ba23)2c3a08e)
</details>6 commits contributed to the release over the course of 32 calendar days.
<csr-read-only-do-not-edit/>
<csr-read-only-do-not-edit/>
<details><summary>view details</summary>
f9fbcba)444a92b)package.include patterns more specific so they don't match ignored files (c2c917f)5f5a836)just nextest work reliably (789b57f)98bae84)
</details>gate experimental APIs with introduction dates
gate experimental APIs with introduction dates
A stable API needs to retain internal access to experimental items while making
those items an explicit opt-in for downstream consumers. Cargo feature unification
also means API checks must distinguish experimental members even when compiled
with the feature enabled.
Add unstable for items, inherent methods and trait implementations, plus
unstable_fields for selected public fields. Disabled gates retain crate-private
access where Rust permits it; trait implementations are omitted instead. Support
file-backed modules on Rust 1.88 through inline bodies with include!;
applying the attribute directly to pub mod name; requires Rust 1.99.
Require since = "YYYY-MM-DD" with the item's introduction date from Git history.
Validate calendar dates and include them in the generated instability marker.
Keep reason optional and nonempty when supplied, for explanations beyond the
signature or the fact that an item is experimental.
Cover visibility, fields, trait implementations, documentation, missing and
malformed dates, leap years and centuries, and invalid or duplicate arguments.
The fixtures use their introduction dates from Git history.
<csr-read-only-do-not-edit/>
<csr-read-only-do-not-edit/>
<details><summary>view details</summary>
c389a2c)6183fd0)6bdb331)f7277f3)
</details>The newly published dua-core 3.3 release used by linked-worktree removal requires Rust 1.88, so raise every workspace crate and the advertised badge t
migrate errors to gix-error
raise MSRV to Rust 1.88
The newly published dua-core 3.3 release used by linked-worktree removal
requires Rust 1.88, so raise every workspace crate and the advertised badge
together.
Keep the MSRV checks buildable by selecting the latest sysinfo and rusqlite
release lines that support Rust 1.88.
From<Message> for ValidationError guide.
This allows to more conveniently create validation errors.gix-error instead of thiserror in gix-quote
Replace the thiserror-derived ansi_c::undo::Error enum with
gix_error::Exn<gix_error::ValidationError>, converting the Error::new()
factory and variant constructors to message!() calls.<csr-read-only-do-not-edit/>
<csr-read-only-do-not-edit/>
<details><summary>view details</summary>
ecf90fc)000d58a)gix-error instead of thiserror in gix-quote (502eaa0)From<Message> for ValidationError guide. (fdf321b)9327b73)
</details>Initial release with pack-verify
pack-verify<csr-read-only-do-not-edit/>
<csr-read-only-do-not-edit/>
<details><summary>view details</summary>
c68d9ab)48c4bbd)1f1b725)cce71aa)f276a05)4aa8022)e2a7bdd)41e01a5)e98e7c2)5fff552)0cc1bf2)aac0d34)af02b46)a4d978c)a1a0b13)5acecc5)1f9bc03)fce7035)b3780f8)13cd8ce)ac490c2)6794300)122d69f)6adf615)5b4979c)7add82c)3bf3321)c416152)60eaea0)caa883b)026a0dd)30925e6)67026e4)aaf4825)b820717)da3ae1c)b1e51d6)0fbba9f)ba6a8ef)f01c298)b09b4e1)0ac9c5a)8c0e0b5)7cec2b6)41f4bce)1179ac1)3ca2c59)aa8efdd)eb7388c)417c34b)87c8a2e)2aa0857)57737c2)c3d319f)
</details>ParseError into ValidationError
The latter is more general and makes sense both for parsing,
and for validation.<csr-read-only-do-not-edit/>
<csr-read-only-do-not-edit/>
<details><summary>view details</summary>
d66ac10)e4f016b)f860c0b)0b0e9f8)9007e1b)8bceefb)7ce3c55)e8612b5)ParseError into ValidationError (b2c516a)f7d0975)
</details>Add an auto-chain-error feature to let gix-error::Error produce error chains suitable for anyhow.
auto-chain-error feature to let gix-error::Error produce error chains suitable for anyhow.anyhow source-chain display.gix-error::Exn
This is mainly useful for gitoxide-core, which may call plumbing.Exn::downcast_any_ref()gix-error in gix-date
This will make for easier introspection for users of these errors.exn.Exn::from_iter to raise_all<csr-read-only-do-not-edit/>
<csr-read-only-do-not-edit/>
<details><summary>view details</summary>
f8517be)829393a)fa302a1)9d39656)16327ef)auto-chain-error feature to let gix-error::Error produce error chains suitable for anyhow. (461c876)anyhow source-chain display. (28f4211)gix-error::Exn (053c3ee)6cff657)e112cac)ErrorExt::erased() to ErrorExt::raise_erased(). (373fced)Exn work properly after the type was erased. (499402c)Exn::downcast_any_ref() (3301eb8)25233ce)Exn::into_box() to Exn::into_inner(). (939b8fc)4c6a7a7)gix-chunk (e6e90ff)ParseError for less noisy invocations. (08f9ed4)gix-error (f46ca99)ae23762)3bac149)gix-error into gix-revision. (4819ea8)exn to most pressing needs of gitoxide (abedade)exn from https://github.com/fast/exn@bb4d8ea4e4df335c46d4fa3f4f260121f9f84305 (0eaab70)gix-error in gix-date (5ab19a7)gix-error crate to forward exn (35cf1ff)
</details>One column per month.