Error messages in mfer/ and internal/cli/ are lowercase except names and
acronyms, carry no "failed to" or command-name prefix, and each wrap names
only the operation and thing the wrapped error does not already name, so a
stacked message names what failed once. Wraps around errors that already
name their operation and path (os and afero path errors, url.Error, the gpg
helpers' own errors) are dropped. runGPG appends gpg's stderr only when gpg
wrote some, so no message ends in a colon. errHTTPStatus reads "unexpected
HTTP status"; both inner-not-set sentinels read "inner message not set".
No sentinel, errors.Is result or exit status changes. Message tests pin the
new text through the real call sites.
Model: opus-5-5
gen given arguments whose files share a path, such as gen a b with a.txt
in both, or gen . ., now fails while listing the files, before hashing
any, naming the path and both files. Builder.AddFile and
Builder.AddFileWithHash refuse a path already added. Loading refuses a
manifest that lists a path twice, compared byte for byte; fetch keeps
its own letter-case check. The Path Rules in docs/FORMAT.md say each
path appears at most once. The decode-size test listed one path 1000
times; each entry now has its own path of the same length.
Model: opus-5-5
Loading a manifest checked only the outer message's version, so an
inner message of version 0 or a later version loaded as if it were
version one, although docs/FORMAT.md requires VERSION_ONE in both.
deserializeInner now refuses any other inner version with the same
error as an unknown outer version. Two existing tests built inner
messages with no version and expected them to load; they now write
version one.
Model: opus-5-5
NewManifestFromReader reads at most one byte past MaxManifestSize, a new
constant of 258 MiB: the 256 MiB decompressed limit grown by zstd's worst
case of 1/256, plus 1 MiB for the signature, the signing key and the
other outer fields. It refuses a larger manifest. fetch, and check given
a URL, stop downloading a manifest one byte past the same size and report
it as too large; tests lower that size to keep their memory small. fetch
stops reading a file one byte past its listed size, so a longer body ends
in the size mismatch at once instead of filling the disk. docs/FORMAT.md
states the limit and gives the decompressed limit as 256 MiB, the size
the code uses.
Model: opus-5-5
check and fetch --require-signature compared the required fingerprint
with the first key in the manifest's embedded public key block, while
gpg accepted a good signature by any key in that block.
Loading a signed manifest now refuses one whose embedded block holds
more than one primary key, counted as gpg reads the block, or whose
signer field is not the primary key fingerprint gpg reports for the
signature. --require-signature compares with the signer field, which
loading has checked. Signing names and embeds the key gpg reports it
signed with, so a key ID matching several keys still writes a manifest
that loads. docs/FORMAT.md states what a verifier checks.
Model: opus-5-5
Go 1.27.1 in go.mod and in the Dockerfile's test and build images.
Every module go.mod requires is at its current release; protoc-gen-go
follows protobuf to v1.36.12 and mf.pb.go is regenerated. The standard
library uuid package replaces github.com/google/uuid; FromBytes could
only fail on a length validateUUID already checks, so that call and its
unreachable error are gone. make vulncheck runs govulncheck v1.8.0,
installed with go install at its release commit, in a vulncheck stage
of the Dockerfile; script/check does not run it. The newer go directive
switches on lint checks for strings.SplitSeq and t.Chdir, now used. A
new test pins the bytes of a seeded manifest written by an mfer built
before this change.
Model: opus-5-5
MFFilePath gains mode (field 304): a file's permission bits, 0777 at
most, or 0000, meaning none recorded. gen and freshen record real modes
only with --include-permissions (ScannerOptions.IncludePermissions); the
builder keeps only mode.Perm(), so setuid, setgid and sticky are never
written. list -l and export show the mode in octal. check reports
MODE_MISMATCH for a recorded mode other than 0000 the file lacks. fetch
refuses a manifest with a mode above 0777 before requesting any file,
sets only the permission bits of each recorded mode on the files it
writes, and downloads again a present file whose mode differs. The decoding-cost bound counts a file
entry at 176 bytes, up from 160.
Model: opus-5-5
atime left mf.proto earlier and nothing reads or writes it, but
docs/FORMAT.md still narrated its removal and listed it among the
determinism rules. The spec now describes the file entry by its fields
only.
A new test compares the MFFilePath message descriptor, by name and
number, with a list copied from the spec's field table. It does not read
the spec, so changing a field means changing the proto, the spec and
that list together.
Model: opus-5-5
gofumpt and protoc-gen-go are tools of a separate Go module in bin/tools,
so `go tool` builds them from source checked against bin/tools/go.sum and
mfer's own module gains no dependencies. script/bootstrap downloads that
module, and unpacks protoc 33.4 into bin/protoc from its release archive
after checking the sha256 it holds for the platform. script/generate runs
that protoc with the pinned plugin, so mfer/mf.go and its go:generate line
go. script/prettier runs only the node_modules prettier, fails when it
differs from the package.json pin, finds the node bootstrap installed
through nvm by the version in .nvmrc, and runs prettier once. Comment and
package.json fixes.
Model: opus-5-5
check now always looks under the base directory for files the manifest
does not list, hidden files and directories included, and prints one
warning per file. The result still depends only on the listed files;
--no-extra-files turns each unlisted file into a failure, and --quiet
hides the warnings but not the failures. A directory that cannot be
listed is reported the same way, and the search goes on past it.
The manifest itself is left out by file identity, as gen and freshen
do; a symlink under the base is compared by what it points to. A base
directory named through a symlink is resolved before the search.
Model: opus-5-5
mf.proto named git.eeqj.de/sneak/mfer/mfer as its Go package, which
disagrees with the sneak.berlin/go/mfer module path go.mod declares.
go_package is now sneak.berlin/go/mfer/mfer, the import path of the
mfer/ package. mf.pb.go is regenerated with make generate, which
changes only the go_package bytes in its embedded descriptor, and
mf.proto.sha256 records the new mf.proto hash.
Model: opus-5-5
AddFileWithHash took any non-empty bytes as a hash, so the builder could
write a manifest that mfer refuses to load. It now decodes the hash with
go-multihash and also requires a digest of at least 32 bytes, the SHA-256
length the reader's decoding-cost limit assumes: a valid but shorter
multihash, such as an empty identity hash or SHA-1, still makes a
manifest of one-character paths too costly to load. When mfer freshen
meets such a hash in an existing manifest, its error names the manifest
entry and says to regenerate the manifest with mfer generate. Test
fixtures whose stand-in hashes were not valid multihashes now use a
SHA-256 multihash.
Model: opus-5-5
mfer gen now writes index.mf instead of .index.mf, the name fetch
requests and the README calls the standard filename. Given a
directory, check, freshen, list and export look only for index.mf;
.index.mf is no longer recognized. Because index.mf is not hidden, gen
and freshen leave out of their own listing the manifest they write and
a temp file an interrupted run left beside it, matched by file
identity (os.SameFile) however the path is spelled. The scanner takes
these as a plain ExcludePaths list; manifestTempPath alone names the
temp file, for both writes, both skips and the tests.
Model: opus-5-5
The test and format scripts regenerated mfer/mf.pb.go whenever
mfer/mf.proto looked newer by mtime, which a fresh checkout often causes,
so make check could rewrite a committed file and needed protoc. Nothing
regenerates it any more except make generate (script/generate), which
refuses to run unless protoc 33.4 and protoc-gen-go v1.36.11, the
versions that wrote the committed file, are on PATH, and records the
hash of mf.proto in mfer/mf.proto.sha256. A Go test compares that hash
with mf.proto and fails, naming make generate, when they differ. The
mtime rule, the unused protoc-gen-go v1.28.1 install rule, make clean's
deletion of mf.pb.go and the Dockerfile's touch workarounds are removed.
Model: opus-5-5
NewChecker now takes *CheckerOptions (ManifestPath, BasePath, Fs), named
like ScannerOptions. A nil Fs still means the OS filesystem, as before and
as in ScannerOptions; nil options or an empty path return an error naming
the missing path.
Audit of the other exported constructors in mfer: NewManifestFromFile took
a filesystem and a path positionally; it now takes
*ManifestFromFileOptions (Path, Fs) with the same nil and empty rules.
NewBuilder and NewScanner take no arguments, NewScannerWithOptions already
takes options, and NewManifestFromReader takes one reader, which the style
guide exempts; these are unchanged.
Model: opus-5-5
Before decoding the manifest, the parser adds up what decoding sets
aside for each file entry, hash, timestamp and MIME type, however short
its encoding, and refuses the manifest once that passes 8 times the
decompressed size; manifests mfer writes come to at most about 7.15
times. Empty entries decoded to about 50 times their size: under 1 KB of
manifest allocated about 500 MB. Fields the decoder does not know are
dropped; kept, they took up to 5 times more. A test refuses entries
counted just over 8 times and loads them just under. The fuzz ceiling
rises from 16 to 20 times the input and decompressed data; seeds of
empty entries and of empty hashes fail it without the fix.
Model: opus-5-5
The two status-send tests now call the send directly on a channel nobody
receives from, so a blocking send hangs the test into its timeout instead
of racing a 100 ms timer that a loaded host can miss.
The gpg test for a child holding gpg's output had the same problem: its
100 ms deadline could fire before the fake gpg wrote the PID of sleep,
and the cleanup then failed. The fake gpg now writes that PID to a named
pipe, and the test cancels only after reading it. It then waits up to
10 s for the call, so a call that waits for sleep fails the test and the
cleanup still kills sleep.
Model: opus-5-5
FuzzNewManifestFromReader fails when the parser returns both or neither
of a manifest and an error, or allocates more than a fixed multiple of
its input and the decompressed data it may read, plus room for the
decoder's window buffers. make test runs the seed corpus; make fuzz
fuzzes for one minute.
Parser bug: MaxDecompressedSize did not bound decompression; the zstd
decoder decoded a payload under 128 KiB in full before the LimitReader
read any of it. It now decodes only what the LimitReader reads and
refuses windows over the 8 MiB mfer writes with. Seeds: a frame claiming
8 GiB, two frames together over the limit, empty frames with growing
windows.
Model: opus-5-5
Every gpg run now has a one-minute deadline (gpgTimeout) on top of its
caller's context and is killed when either ends. A timeout is reported
as "gpg timed out" under the failing operation instead of "signal:
killed". Only gpg itself is killed; WaitDelay (one second) stops the run
from waiting on a process gpg left behind that still holds its output,
such as a wrapper script that does not exec the real gpg.
Builder.Build and Checker.ExtractEmbeddedSigningKeyFP take a context, so
ToManifest's context now reaches signing and the contextcheck
suppression calling signing non-cancellable is gone. Manifest loading
takes no context, so its signature check is bounded by the timeout
alone.
Model: opus-5-5
Untrusted .mf files were parsed with no path validation, so an entry like ../../etc/passwd reached filepath.Join against the checker's base path and mfer check could stat and read outside it. ValidatePath ran only when building a manifest. It now runs on every entry as the manifest loads, so every consumer is covered. The whole manifest is rejected on the first bad entry instead of dropping it, which could hide files from a check; the error wraps errInvalidManifestPath and names the path.
Disclosure: a path that is not valid UTF-8 is refused earlier, by the protobuf string decoder, so that error does not name the path.
Model: opus-4-8 (implementation); fable-5-1 (summary)
Adopts golangci-lint v2.12.2 and the canonical .golangci.yml (default: all), and fixes all resulting findings across the tree.
Two intended behavior changes: absent MFFilePath.Mtime is handled explicitly in freshen, list and export rather than dereferenced (main panicked); gpg positional key IDs now follow an explicit -- end-of-options marker.
All twelve reworded user-visible error messages restored to byte-identical parity with main and pinned by tests.
Replace 150M SHA-256 iteration key-stretching with a single hash.
Remove all references to iteration counts, timing (~5-10s), and
key-stretching from code and documentation.
The seed flag is retained for deterministic UUID generation, but
now derives the UUID with a single SHA-256 hash instead of the
unnecessary iterative approach.
Adds a --seed CLI flag to 'generate' that derives a deterministic UUID
from the seed value by hashing it 1,000,000,000 times with SHA-256.
This makes manifest generation fully reproducible when the same seed
and input files are provided.
- Builder.SetSeed(seed) method for programmatic use
- deriveSeedUUID() extracted for testability
- MFER_SEED env var also supported
- Test with reduced iteration count for speed
Sort file entries by path (lexicographic, byte-order) before
serialization to ensure deterministic output. Add fixedUUID support
for testing reproducibility, and a test asserting byte-identical
output from two runs with the same input.
Closes#23
FindExtraFiles now skips hidden files/directories and manifest files
(index.mf, .index.mf) when looking for extra files. Previously it would
report these as 'extra' even though they are intentionally excluded from
manifests by default, making --no-extra-files unusable.
Also includes IsHiddenPath fix for '.' (needed by the new filtering).
path.Clean(".") returns "." which starts with a dot, causing IsHiddenPath
to incorrectly treat the current directory as hidden. Add explicit checks
for "." and "/" before the dot-prefix check.
Fixed in both mfer/scanner.go and internal/scanner/scanner.go.
- Implement gpgVerify function that creates a temporary keyring to verify
detached signatures against embedded public keys
- Signature verification happens during deserialization after hash
validation but before decompression
- Extract signatureString() as a method on manifest for generating the
canonical signature string (MAGIC-UUID-MULTIHASH)
- Add --require-signature flag to check command to mandate signature from
a specific GPG key ID
- Expose IsSigned() and Signer() methods on Checker for signature status
- Add UUID field to both inner and outer manifest messages
- Generate random v4 UUID when creating manifest
- Hash compressed data (not uncompressed) for integrity check
- Verify hash before decompression to prevent malicious payloads
- Validate UUIDs are proper format and match between inner/outer
- Sign string format: MAGIC-UUID-MULTIHASH
- Add --sign-key flag and MFER_SIGN_KEY env var to gen and freshen commands
- Sign inner message multihash with GPG detached signature
- Include signer fingerprint and public key in outer wrapper
- Add comprehensive tests with temporary GPG keyring
- Increase test timeout to 10s for GPG key generation
- Add FileCount, FileSize, RelFilePath, AbsFilePath, ModTime, Multihash types
- Add UnixSeconds and UnixNanos types for timestamp handling
- Add URL types (ManifestURL, FileURL, BaseURL) with safe path joining
- Consolidate scanner package into mfer package
- Update checker to use custom types in Result and CheckStatus
- Add ModTime.Timestamp() method for protobuf conversion
- Update all tests to use proper custom types
Use the FileInfo already provided by Walk instead of calling Stat again.
Only stat if fi is nil (defensive, shouldn't happen in normal Walk usage).
Also fixes potential nil pointer dereference if fi was nil.
sneak
clawbot
clawbot