Skip to content

Pull request review — 2026-09-14

Historical record. This preserves the source, tests and limitations recorded at the time; it is not current setup guidance. See the documentation index.

Scope: integrate PR #86 with current runtime behavior, then review all open PRs, retaining useful work from anagnorisis2peripeteia and closing stale or superseded proposals with evidence. No released tag or asset is replaced by this source work.

Review decisions (not a record of completed GitHub actions)

PR Assessment
#86 Useful DSX mod compatibility feature; original patch requires protocol, lifecycle, reversible output ownership, and settings corrections before integration.
#41 Retain the authored non-default AppSettings deserialization test, without merging unrelated history from its stale branch.
#42 Production invariant-date parsing/formatting is already superseded by stronger current code. Retain the remaining culture-independent expected-date assertion.
#43 Retain macro value-band, boundary, idempotence, and output-name tests; assert the expected count too.
#44 Do not merge: removed gyro fields are now used by reference in active mouse mapping.
#45 Retain profile migration v1–v5 coverage.
#47 Superseded by current independent RGB channel parsing and invalid/missing/precedence tests.
#48 Do not merge stale instructions that exclude repaired serialization tests or claim CI x86 coverage not present in the current workflow.
#49 Superseded by current shifted-extras serialization and stronger extras-only activation/release handling. Its proposed save fixture omits the required positive shift trigger.
#7 / #8 Do not blindly upgrade MSTest to v4: the repository still uses hundreds of Assert.ThrowsException calls removed by v4. Framework-only upgrading also mismatches the adapter. Requires coordinated migration.
#10 System.Management 10.0.8 supports net8, but changes the actual WMI runtime and System.CodeDom dependency. Requires application and installer-helper WMI validation; not merely a version edit.
#11 System.Memory 4.6.3 is a plausible low-risk dependency update; still requires restore/build/test validation before integration.
#70 Preserve native UdeCx experimental work; do not merge into production. It targets viiper-ui-rebuild, carries local-test-only metadata, and requires protected production driver signing/package evidence.

The source review of #70 covers production packaging/runtime boundaries, not a complete security audit of the experimental driver.

Completed integration and disposition

  • #86 merged in d760391; author ancestry retained, with the integration corrections and documented compatibility boundaries below.
  • #43 merged in 7382275: five additional macro tests, plus an expected-count assertion preventing an empty/empty string-generation false pass.
  • #45 merged in 61593f9: six additional historical profile migration tests.
  • #41 incorporated and closed: authored test cherry-picked as 611e805, preserving Cameron Beeley's authorship without unrelated branch history.
  • #42 incorporated/superseded and closed: remaining date assertion retained in 19e53de with contributor credit; stronger production parsing preserved.
  • #44, #47, #48, #49 closed with individual explanations and evidence for the obsolete or superseded changes. Contributor branches were not deleted.
  • #7, #8, #10, #11, #70 remain open for the separate migration, validation, and production-driver requirements identified above. None was blindly merged.

GitHub confirms the three merges and six closures. All source changes are pushed to main. No released tag, public release asset, installed application, driver, controller session, startup task, or registry setting was changed by this review.

Final combined validation after the follow-up integrations: 6,362 passed, zero failed, 11 gated skips (6,373 total), x64 Release. Focused macro tests: 7/7; profile migration: 8/8; settings/profile/date: 30/30. Evidence: _results/pr-followups/all-prs-final.trx, pr43.trx, pr45.trx, and pr41-pr42.trx in that same results directory. Hosted CI is separate from these local results; this document does not claim a new release or physical game-mod acceptance.

Primary references

PR #86 integration corrections

  • Correct wire trigger IDs, player LED enumeration, mic pulse, RGB brightness, and frequency-first multi-position vibration layout.
  • Strict bounded complete-packet validation; reject invalid indices/parameters rather than targeting slot 0, coercing values, or inventing effects.
  • One bounded response per accepted datagram, including mutation-only packets.
  • Session-owned local sockets, generation-safe callback admission, and reset/stop/disconnect cleanup without profile mutation.
  • Per-field physical output overlays and per-trigger Trigger Lab priority. Preserve exact native effect bytes for later restoration.
  • Default off, loopback-only, editable port, retry after bind failure, and actual listener status distinct from the saved preference.
  • Remove the unrelated obsolete VIIPER 0.1.2 checksum. Replace the original automatic strong-effect script with status-only default and explicit bounded effect testing that requests reset in finally.

Hardware mod acceptance remains separate from automated parser, loopback, ownership, FIFO, and canonical output-composition tests. No physical effect test, live controller restart, installer action, or reboot is part of this audit.

Combined focused validation: 517 passed, zero failed, x64 Release net8.0. This includes ephemeral loopback traffic, official legacy RGB/custom PulseAB/null-padding packets, exact wire vectors, malformed-packet rejection, stop/restart/reentrant callbacks, occupied-port retry, unexpected listener exit, actual service routing with fake controllers, replacement ownership, native admission, USB/BT composition, reset ordering, and Trigger Lab priority.

Review also corrected repeated trigger strobes, stale local trigger proof after native/reset intervention, and early publication of unadmitted native Trigger Lab bytes. Exact native commands remain ordered; no new input mapping stack, unbounded output queue, or physical writer was added.

The unchanged warmed native-shadow allocation assertion measured zero bytes over 6,000 observations, after fixing value-type equality that initially boxed 256 bytes per report. The existing deliberate-allocation positive control also passed. This is not a hardware effect-quality or end-to-end latency result.

Full x64 Release suite: 6,350 passed, zero failed, 11 gated skips (6,361 total). The first full run identified a source-contract test still searching for the old void return signature; its markers were updated to byte, preserving all fixed-scratch/no-per-report-allocation assertions. The entire suite was then rerun successfully. Evidence: _results/pr86-full-final/pr86-full-final-rerun.trx. The installed-cache fixture used a read-only protected bundle snapshot; no setup or controller operation ran on this PC.