BT411

Files

T

Joe DiPrimaandClaude Opus 5 45750a6fcf SENDLOGS.bat: fix the two field failures + stop it lying about success. Two testers hit DIFFERENT errors, both now reproduced on a rig and fixed.

(1) ConstrainedLanguage mode (locked-down Windows, WDAC/AppLocker) blocks .NET method calls, so [Environment]::GetFolderPath('Desktop') returned nothing -> null DestinationPath -> no zip, and [math]::Round blew up on the size. Desktop is now resolved from OneDrive/USERPROFILE env vars (still honouring a redirected Desktop) and the size via -f formatting. Deeper trap found while testing: Compress-Archive ITSELF cannot run in that mode -- it is a script module that calls .NET internally -- so fixing the path alone would only have moved the failure downstream. The zip is now built with Windows' native tar.exe (immune to language mode, present since Win10 1803, verified to emit a real Explorer-readable zip), with Compress-Archive kept only as a fallback.

(2) A log still held open by a running BattleTech made Compress-Archive abort the ENTIRE zip -- one locked file, no logs at all. Files are now staged with Copy-Item first: an unreadable one is skipped and NAMED, the zip still gets made from the rest, and the tester is told to close the game and re-run to capture the missing one.

(3) The script announced 'Wrote <path>' and 'Attach the BTLOGS zip' unconditionally -- tester 2's screenshot shows it claiming '11 files, 0 MB' for a zip that was never created. Success is now gated on Test-Path, failure exits 1, and the batch prints 'Nothing was created' instead of sending people to hunt for a file that does not exist.

Also adds ABOUT.txt inside the zip (machine, pack time, install folder, exe build date/version) so a report can be tied to a BUILD -- the ambiguity that has cost us repeatedly when a Pass/Fail arrives with no build number.

Verified on four scenarios: normal run, a log locked by another process, a session downgraded to ConstrainedLanguage (reproducing tester 1's exact MethodInvocationNotSupportedInConstrainedLanguage error first), and running from a folder with no content dir. Handout updated: close the game first, and what SKIPPED means.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

2026-08-15 01:10:18 -05:00

feedback

Oracle's weapons-guide return ingested -- the file-edit feedback loop WORKS (first curated .txt return, staged to docs/feedback/ + diffed against the distributed bb4094f copy). Fourteen answers: two CONTRADICTS filed (#176 seek-gear damage swing -- the authored seekVoltage gear table has never been read and may reconcile both; #177 veteran heat -- witness + 4.0 manual say expert-only and the player+0x260 writer is unread [T4]). Confirmed by witness: the out-of-range damage cliff, gear-wrap safety + downshift-for-rate doctrine, charge persistence across generator reassignment (port audit item), splash, SRM 800 pip (450/800 CLOSED), single-tap jam eject, missile speed-drop pre-endorsed. New era gifts: the PPC MFD collapse-to-vertical-line reference photo (ARCHIVED before the Discord link expires -- the visual target for Cyd's ppc-sync pass), PPC auto-mode generator-flip color, AFC no-visible-projectile memory (needs corroboration -- our tracer code is real, its VISIBILITY on the shipped machine is now an era question), the Gauss carrier's variant name (Mischief), and two YouTube videos with the real incoming alarm audible (#83 ear-A/B references).

2026-08-13 22:53:08 -05:00

ASSET_PIPELINE.md

Initial commit: bt411 -- standalone Windows BattleTech (Tesla 4.10 port)

2026-07-05 21:03:40 -05:00

AUDIO_FIDELITY.md

KB: #83 paperwork -- gotcha 33 (a fix wired into a blocked transcription: compiles, reads faithfully, never runs; rules: fixes land in the LIVE path, inert transcriptions get an explicit banner, a doc-header FIXED is a claim not evidence, the bench must fail first). decomp-reference section 5 gains the recovered alarm rule (@004bf5b4 ctor socket-registration into target+0x418, @004a9b5c first-member read, death-path list empty, reset 0/FLT_MAX) + env rows BT_INCOMING_LOG / BT_DESIGNATE. AUDIO_FIDELITY F7 corrected: the July fix was dead code; the true fix + bench numbers recorded, field-verify Friday. Tracker #83 comment 1786.

2026-08-13 22:28:22 -05:00

BGF_FORMAT.md