Skip to content

Docs: newer trap and failure registers; the stock DELAY reads its whole table run - #1229

Merged
repeat98 merged 2 commits into
mainfrom
claude/docs-import-54200264
Nov 5, 2026
Merged

repeat98 merged 2 commits into
mainfrom
claude/docs-import-54200264

Conversation

@repeat98

@repeat98 repeat98 commented Nov 5, 2026

Copy link
Copy Markdown
Owner

The vendored trap and failure registers move to the upstream revision of 5 November 2026, byte for byte, in an eighth docs-only import record. The main correction for Modwerk: the stock DELAY reads its coefficient table across the whole zero run 0x400d24d0..0x400d2ce0, not 16 bytes.

What changes

  • sdk/octabam/docs/TRAPS.md and FAILURE_MODES.md: upstream's files at 54200264, unchanged. HARDWARE_MATRIX.md and MIDI_TRACKS.md are identical upstream and stay pinned by their earlier records.
  • sdk/imports/octabam-docs-54200264.json: the record (source and vendored SHA-256, review, limitations), added to DOC_IMPORTS in sdk/tests/test_infrastructure_regressions.py.
  • sdk/octabam/docs/remixer/PLACEMENT.md (Modwerk's own copy, no record pins it): 0x400d24d0..0x400d2ce0 is no longer listed as a free run. The stock DELAY reads all of it as entries 508..1023 of its table: its low-pass index comes from BASE + WIDTH (up to entry 1016), so the default settings read 0x400d24d0, and boot reads the last word. The 0x400c45b0 run is marked as the zero tail of a glyph block, with whether a character reads it unknown.
  • docs/OCTATRACK_HARDWARE_TRAPS.md: the register lists the three entries that reach Modwerk (the DELAY run; the page plane 0x46c7e0ea lacks windows and popups, which UI captures add through lcd_view.py but a bare peek does not; a file open through the stock layer owns the caller's buffer). The card-write race paragraph notes it was seen again in emulator runs. The file is over the 40 KB limit, so it was trimmed elsewhere and is 2 bytes smaller.
  • docs/OCTABAM_INFRASTRUCTURE.md: the eighth import's section. Changelog entry.

No code, build input, module source or tool pin changes. The card builders' guard for the DELAY run is the separate base-coupled change in #1045.

Review

  • Stock-free scan against the original 1.40C MAIN image (unpacked locally from the stock OS, MAIN 164f3122…; nothing written): TRAPS.md has no hex string of 8 bytes or more; FAILURE_MODES.md has nine, and no 8-byte window of any occurs in MAIN or in its DSP payloads read as big-endian words. The same scan found planted stock bytes in an in-memory probe.
  • No instruction listing or encoding. One thing for the reviewer: the rewritten DELAY entry in FAILURE_MODES.md names the low-pass index computation as three instructions by mnemonic at a stock address. It describes how the index is formed and carries no bytes, like the single instructions by mnemonic an earlier docs record accepted. Its one 32-bit value is a word of a module's page cave, not stock.

Checks

  • sdk/tests/test_infrastructure_regressions.py: 22 tests pass.
  • npm run pr:ready: Ready (full check, sdk:check passed).

What the guides lacked

sdk/octabam/docs/remixer/PLACEMENT.md listed the DELAY's table run as free space, and our trap list said nothing about it; both are fixed here. Nothing else.

🤖 Generated with Claude Code

repeat98 and others added 2 commits November 5, 2026 09:54
…s the whole table run

The vendored TRAPS.md and FAILURE_MODES.md are at upstream 5 Nov 2026 byte
for byte, in an eighth docs-only import record that joins the docs-import
tests. The stock DELAY reads its coefficient table across the whole zero run
0x400d24d0..0x400d2ce0 at default settings and at boot, so PLACEMENT.md no
longer lists it as free, and the hardware traps list it with the page plane
and the stock file layer's buffer (trimmed elsewhere so the file does not grow).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@github-actions github-actions Bot added the documentation Improvements or additions to documentation label Nov 5, 2026
@repeat98
repeat98 merged commit ac19063 into main Nov 5, 2026
9 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

documentation Improvements or additions to documentation

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant