build: depend on helly25_mbo 0.11.1 from BCR (drop git_override) - #45
Merged
Conversation
0.11.0 was pinned to a main commit via git_override because the macOS clang+asan verification (#188) and the StatusIs code-matcher (#190) we rely on were not yet in a release. Both ship in the 0.11.1 BCR release, so the override is no longer needed -- depend on the registry version like every other module. MODULE.bazel.lock already carries the BCR 0.11.1 MODULE.bazel/source.json hashes (no git refs remain), so no lock change is required.
helly25
added a commit
that referenced
this pull request
Jun 28, 2026
…oadmap decisions (#193) Per the directive to put enforcement rules in the repo, not just memory: AGENTS.md gains two sections: - Self-documenting features: the registry + globals are the doc SOT; every feature add/change updates its Descriptor.summary / GlobalFlag entry + kHelpText in the same change, so --help/--man/--markdown stay complete. - CLI conventions: flag scope by dash count (--global vs -primary; -h/-q/-help/ -version are special-cased compat globals); flag-only, no subcommands; and a user-toggleable boolean capability is a --feature, not a one-off flag. TODO.md roadmap decisions (design phase, accounting for existing code): - #43 parallel traversal + --sort: already BUILT (ReadPool, sort modes, tests, tsan cell); remaining is a CLI bashtest. - #45 --exact/--path-encoding: default = filesystem-native (natural per-platform case behavior); --exact forces verbatim byte matching; --path-encoding=raw|escape. - #73 --feature: PARKED (no customer yet) with the full ready-to-build design + a trigger (first boolean capability builds it); trigger mirrored in AGENTS.md. - #54 mode mechanism: subsumed by --config (no --mode flag; --modern deferred).
helly25
added a commit
that referenced
this pull request
Jun 28, 2026
#194) The plain renderer can now C-escape path bytes so a newline or control character in a filename cannot corrupt the line-oriented -print stream: --path-encoding=escape turns `\` and control bytes into `\\`, `\n`, `\t`, `\r`, or `\xNN`; raw (the default) stays verbatim and find-compatible. It applies to the plain format only -- kNul keeps raw bytes (the NUL is the separator) and kJsonl always JSON-escapes. render::Renderer gains a PathEncoding; run.cc ResolvePathEncoding mirrors ResolveFormat (last wins, raw default) and feeds the implicit-print renderer. Self-documented via the globals table + kHelpText (so --help/--man/--markdown list it). render_test covers escape/raw and that it applies only to plain. The other half of #45 (--exact, filesystem-native matching) remains.
helly25
added a commit
that referenced
this pull request
Jun 28, 2026
…rxc (regex) (#195) * cli: add --path-encoding=raw|escape for safe plain output (#45 part 1) The plain renderer can now C-escape path bytes so a newline or control character in a filename cannot corrupt the line-oriented -print stream: --path-encoding=escape turns `\` and control bytes into `\\`, `\n`, `\t`, `\r`, or `\xNN`; raw (the default) stays verbatim and find-compatible. It applies to the plain format only -- kNul keeps raw bytes (the NUL is the separator) and kJsonl always JSON-escapes. render::Renderer gains a PathEncoding; run.cc ResolvePathEncoding mirrors ResolveFormat (last wins, raw default) and feeds the implicit-print renderer. Self-documented via the globals table + kHelpText (so --help/--man/--markdown list it). render_test covers escape/raw and that it applies only to plain. The other half of #45 (--exact, filesystem-native matching) remains. * Add content-search predicates: -content/-icontent (literal) + -rxc/-irxc (regex) File-content search (rg/fd parity): the xff counterpart of -name/-regex, but over a file's bytes instead of its path. -content STR / -icontent STR literal substring (i = ASCII case-fold) -rxc RE / -irxc RE RE2 regex, unanchored (PartialMatch) All four are xff extensions (--config=find rejects them) and Cost::kExpensive (they read the file). Binary files are skipped by default via a NUL-byte sniff over the first 8 KiB (grep/ripgrep behaviour). The name -grep was deliberately rejected: grep's BRE/ERE/-P flavor baggage is ambiguous for both a literal and a single-flavor-regex predicate. Mechanism: - regex::Matcher::PartialMatch (RE2::PartialMatch), backing -rxc/-irxc. - vfs::FileSystem::ReadContent (whole-file read) across all implementers: LocalFs uses a POSIX read loop, DryRunFileSystem delegates, the test fakes stub it. - evaluate.cc ContentToSearch (type-gate + read + binary sniff) + EvalContent/EvalRxc. - parser CompileNodeRegex precompiles -rxc/-irxc like -regex/-iregex. Self-documentation in the same change (registry SOT): four Descriptor summaries (flow into --help / --help=NAME / --man / --markdown), the kHelpText expression summary, and a resolved-names note in docs/design.md superseding the -contains/-grep placeholders. Tests: regex PartialMatch; LocalFs::ReadContent; evaluate_test literal / regex / case-fold / binary-skip / non-regular; content_test.sh end-to-end (incl. find-style gating and the --help topic).
helly25
added a commit
that referenced
this pull request
Jul 2, 2026
Foundation for xff's FS-native name matching: a per-volume case-sensitivity query on the FileSystem interface. LocalFs answers it via pathconf(_PC_CASE_SENSITIVE) on macOS/BSD (1 = case-sensitive, 0 = folding); on Linux, where glibc has no such name, it reports case-sensitive (ext4/xfs/btrfs default; per-directory casefold via statx STATX_ATTR_CASEFOLD is a later refinement). An unprobeable volume falls back to case-sensitive (byte-exact), which is always find-faithful. The engine's DryRunFileSystem forwards it; the walk/run test fakes return case-sensitive. local_fs_test cross-checks the probe against the volume's actual behaviour (whether FILE.TXT resolves to a lower-case file.txt), so it is correct on both case-sensitive and case-folding runners without hard-coding a platform. No behaviour change yet: --exact and the FS-native -name default consume this next.
helly25
added a commit
that referenced
this pull request
Jul 2, 2026
* Add --exact / FS-native name matching (#45) The xff style now matches -name / -path the way the entry's own volume resolves names: on a case-folding volume (APFS / HFS+ default, NTFS, exFAT) a lower-case -name pattern matches a mixed-case file, the way the OS itself would resolve the lookup; on a case-sensitive volume (ext4 and friends) it stays byte-exact. This fixes find's surprising miss on macOS where -name foo.txt skips Foo.txt. --exact opts out, forcing verbatim byte-exact matching regardless of the volume; the find style is always byte-exact (drop-in faithful), as is the conservative in-process (nullopt-style) default. The -iname / -ipath variants fold regardless; regex keeps its own -iregex. Mechanics: run.cc resolves the FS-native policy once (xff style and not --exact), then per entry consults vfs::IsCaseSensitive cached by device id (single-threaded visitor, no lock), setting EvalContext::fold_name_case for a folding volume; a probe miss/error falls back to case-sensitive. EvalName/EvalPath OR that into the descriptor's fold_case. Self-doc: --exact in globals.cc + kHelpText (Matching). design.md reconciles the older tri-state sketch to the shipped boolean (case only; NFC/NFD + fuzzy stay deferred); TODO.md #45 marked shipped. Tests: evaluate_test (fold_name_case folds -name/-path, -iname still folds); run_test (CaseFoldFs: xff folds, --exact / find / nullopt stay byte-exact, exact-case always matches); exact_test.sh end-to-end, volume-probed so it holds on case-sensitive and folding runners. bazel test //... green (37). * shfmt: single space before trailing comments in exact_test.sh
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
helly25_mbowas pinned to amaincommit viagit_overridebecause the fixes our test conventions rely on — the macOS clang+asan verification (#188) and theStatusIscode-matcher support (#190) — were unreleased at the time.Both ship in mbo 0.11.1, which is published in the Bazel Central Registry. So the override is now pure cruft: depend on the registry version like every other module (
abseil-cpp,re2,googletest, …).MODULE.bazel.lockalready carries the BCRhelly25_mbo/0.11.1MODULE.bazel+source.jsonhashes and no longer references any git ref;bazel mod graph --lockfile_mode=updateleaves it untouched, so no lock change is needed.Test plan
bazel mod graphresolveshelly25_mbo@0.11.1from BCR with no errors.bazel test //...(default) and--config=clang --config=asan— including thefieldsconstexprLimitedMapdispatch that uses mbo's container.