* fix(esp): disclose partition entries outside the table checksum
* fix(esp): label the partition checksum offset origin
* fix(esp): count partition entries outside the table MD5
The coverage warning was only prose, and it presented the MD5 entry's
table-relative offset without a structured fact. Record
entriesAfterChecksum on the MD5 result, project it as
entries_after_checksum, and keep the warning as presentation.
Co-authored-by: morluto <morluto@users.noreply.github.com>
* fix(esp): accept saved flash inventories without the checksum count
A required entries_after_checksum rejected ESP flash Evidence written
before the field existed. Keep the count exactOptional so omission
stays unknown, and still emit it for new inventories.
Co-authored-by: morluto <morluto@users.noreply.github.com>
---------
Co-authored-by: ShuaixinHuang <197200901+ShuaixinHuang@users.noreply.github.com>
Co-authored-by: ShuaixinHuang <ShuaixinHuang@users.noreply.github.com>
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
Co-authored-by: morluto <morluto@users.noreply.github.com>
* fix(desktop): bind dconf profiles to selected database sources
* fix(desktop): resolve dconf before changing working directory
* fix(desktop): retain refused dconf lookup diagnostics
* fix(desktop): preserve execution-time dconf search fallback
* fix(desktop): keep host spawn failures outside dconf fallback
* fix(desktop): retain staged reader cleanup through command close
* test(kuna): await eviction boundary events explicitly
* test(kuna): bound lifecycle acknowledgements through shared owner
Cherry-picked shared follow-up from a5761c7545.
* fix(desktop): keep dconf sources independent of wrapper stdin
* test(desktop): restrict procfs command checks to Linux
* fix(desktop): own dconf wrapper cleanup before source retirement
* fix(desktop): retain dconf preparation inspection owners
* fix(desktop): classify dconf read failures by kind and errno
Record the failure kind, host errno, exit status, selected command, and
child stderr on DconfDump, and carry those facts to view_failure. Tests
assert the typed fields instead of error prose, including a PATH refusal
whose executable is not the last candidate.
* test(desktop): gate real dconf diagnostics on Linux
The real reader checks /proc file descriptors before launch, so macOS
reports descriptor-unavailable/ENOENT instead of launch-failed/EACCES.
Keep that launch refusal on Linux and cover the Evidence contract with
a mocked reader on every platform.
Co-authored-by: morluto <morluto@users.noreply.github.com>
---------
Co-authored-by: ShuaixinHuang <197200901+ShuaixinHuang@users.noreply.github.com>
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
Co-authored-by: morluto <morluto@users.noreply.github.com>
* fix(node-sea): bound duplicate blob verification work
* test(mcp): observe cleanup failures at their current delivery boundary
* fix(mcp): preserve workflow completion progress through delivery
* test(setup): keep the Windows Kimi profile fixture on a supported profile
The Windows-only case at `clientConfigurationEnvironment.test.ts:226` wrote
`rea_mcp_input_schema_profile: "full"`. That profile was retired: the reader
now refuses it ("REA_MCP_INPUT_SCHEMA_PROFILE accepts only \"compact\" …"),
so `configureClientConfiguration` returned `failed/readback` before writing
and the case failed on every Windows run, including main's own
`Windows capability and package lane`.
The case has no Windows dependency — it passes an explicit `configPath` and
the Kimi write dialect is the same everywhere — so it now runs on every
platform, which is what let the stale value rot unnoticed. The fixture keeps
the lowercase explicit key the case is about, with the supported value.
Verified on `main` `d6bcfc10`: the same fixture driven through
`configureClientConfiguration(..., "win32")` on this host reproduced
`{"status":"failed","reason":"readback"}` with the retired value and now
returns `{"status":"configured"}` with `compact`. The file's 75 cases pass
here (66 run, 9 platform-conditioned skips) and `check:fast` passes.
* test(ci): restore verifier and MCP contract regressions
* test(captures): assert adapter-owned input issue field names
* fix(ghidra): avoid Java probes for missing installations
* test(ghidra): drop duplicate absent-installation coverage
Merging main reapplied the #11732 case beside the copy already on main.
Co-authored-by: morluto <morluto@users.noreply.github.com>
* test(node-sea): assert duplicate-blob refusal without stdout prose
The CLI coverage already checks the truncated code and limit details.
Co-authored-by: morluto <morluto@users.noreply.github.com>
---------
Co-authored-by: ShuaixinHuang <197200901+ShuaixinHuang@users.noreply.github.com>
Co-authored-by: g0at <50488370+g1at@users.noreply.github.com>
Co-authored-by: IvenKooLab <276572622@qq.com>
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
Co-authored-by: morluto <morluto@users.noreply.github.com>
A LinkInfoHeaderSize below 0x1C no longer lowers the string-offset floor.
Reserved 0x1D..0x23 headers still end at their declared size, so ANSI
fields after that header stay readable.
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
Co-authored-by: morluto <morluto@users.noreply.github.com>
* ci: retry bounded Ubuntu archive updates
* ci: retry MinGW package downloads
* fix(ci): bound Ubuntu toolchain acquisition end to end
* fix(ci): fall back to the Ubuntu Azure archive for Noble layout
---------
Co-authored-by: morluto <76467478+morluto@users.noreply.github.com>
* fix(artifacts): preserve ASAR unpacked path identity
* test(artifacts): use the current ASAR resource entry
* test(artifacts): assert ASAR unavailable content facts
---------
Co-authored-by: ShuaixinHuang <197200901+ShuaixinHuang@users.noreply.github.com>
* fix(environment): map id-list elements when a scalar set also changed
When comparing an answer's scalar id list, the id mapping ran only on the
sameItems fast path. Once one element genuinely changed, or the lengths
differed, array fell to pairwise comparison that compared each element
without its id-list field, so scalar skipped the mapping: comparing live
[101,102] with env [1,9] under 101->1, 102->2 reported /ids/0 (101 vs 1) as
a change as well as the real /ids/1 (102 vs 9).
Thread an idList flag from array to scalar, set by the same per-element rule
sameItems uses, so a mapped element matches and is counted as a normalised
id while a genuine change is reported with its original value. Non-id arrays
are unchanged.
* fix(environment): pair id-list survivors by translated identity
The first fix mapped each id-list element but still paired scalar arrays by
position, so a survivor that shifted position paired with the wrong item:
[101,102] versus [2] under 101->1, 102->2 still emitted a false /ids/0 value
difference even though 102->2 is the survivor and only the removal changed.
Pair id-list survivors by the environment identity they map to, then any
remaining live and environment items by position. The translation is one owner
shared with sameItems, and the paired comparison still maps the original live
id and reports its original value on a genuine change. Non-id arrays keep
positional pairing.
* fix(environment): keep id-list order checks to matched survivors
#idListPairs appended a positional fallback pair for a replaced id, and the
order check read every pair, so a single surviving identity that shifted
emitted a spurious order difference: [101,102] versus [2,9] under 101->1,
102->2 reported order as well as the real /ids/0 value change.
Return the identity-matched survivor pairs separately and compute order from
them alone; the positional fallbacks are value replacements, not the same
item in a new position. Two or more reordered survivors still report order.
* perf(environment): limit ID recovery to conflicting components
* fix(environment): prove response list order across maximum matches
* fix(environment): detect complete duplicate ID list reorders
---------
Co-authored-by: morluto <76467478+morluto@users.noreply.github.com>
Keep DTB and FIT observations when an admitted alignment gap contains
nonzero bytes. Retain one parser-owned diagnostic with the count and
first file offset.
Refs #10179.
Co-authored-by: Nianze Wu <wunianze666@gmail.com>