Skip to content

PRM intake: the document owner's answers and the November 2026 update - #14

Merged
WDrijver merged 2 commits into
mainfrom
prm-intake-answers
Nov 3, 2026
Merged

WDrijver merged 2 commits into
mainfrom
prm-intake-answers

Conversation

@WDrijver

@WDrijver WDrijver commented Nov 3, 2026 •

Copy link
Copy Markdown
Contributor

Tommo, the PRM's document owner, answered the questions from the intake (PR #7) on 2026-11-03 and sent an updated ODT (update nov 2026). This PR replaces sources/official/AC68080PRM.odt, changes the converter (tools/intake/prm68080.py) and regenerates the files.

Changes

Empty encoding boxes. The original leaves some boxes unmerged so the field name is centered. They are drawn without a border towards their field. The converter now reads the cell borders: an empty box with no border towards exactly one named field becomes part of that field.

Instructions Before After
Bcc.S+, BRA.S+, BSR.S+ bit 7 empty, displacement 6–1 displacement 7–1, bit 0 = 1
MOVE2, MOVEH, MOVEX, MOVZ2 bits 15 and 12 empty, B 14–13 B 15–12
BFLY, TRANS, UNPACK1632 bit 11 empty, d 10–9 d 11–9
DBcc.L, FDBcc.L bit 15 empty, displacement 14–1 displacement 15–1, bit 0 = 1

The last two rows were not in the intake questions. The same border rule finds them. Only FMOVEM still shows empty boxes: whole long words left blank in the memory format.

Programming model. PC is 32 bits. CC is the 8-bit condition-code part of the 16-bit SR, and SR is not part of the user programming model. Before, the table gave the drawn width (64 bits) for both.

Gray text. The reading guide in index.md now says what gray means in the original: less relevant, removed or unused.

PNOT. peor.md gets the rule: PNOT is not an instruction; use peor.w #-1,d1,d2, as the page's note shows.

Status

18 files are back at draft, because the converter does that for any changed body: 15 from the converter changes, plus PCMPccB, STOREILM and MOVEC from the new ODT. The 18 need a new review before they go back to reviewed. The other 71 stay reviewed; only their source line changes to "update nov 2026".

From the new ODT

  • Typos Tommo confirmed:
    • MOVZ2 example: 02 23 45 67 is now 01 23 45 67;
    • FDBcc operation: PC – 1 → Dn is now Dn – 1 → Dn;
    • PCMPccB page: bsel d2,d1,e0 is now bsel d1,d2,e0.
  • MOVEH and STOREILM now say "where a1/a2 points to", matching their examples.
  • MOVEC gets the note "Greyed out are not present in recent cores".
  • The B-field cells of MOVE2, MOVEH, MOVEX and MOVZ2 are merged. The output is the same as with the border rule above.

Converter: styles by properties. LibreOffice renamed the code style P47 to P59 in this version, which broke the code blocks on the overview pages (integer-unit, fpu-unit, optimizing). The converter now recognizes the styles it needs (instruction name, subheading, cover tool lines, code) by a hash of their properties. It leaves out attributes without a visible effect (opacity, use-window-font-color). In both ODT versions each role matches exactly the intended styles, and with the old ODT the converter still gives the committed output.

Still open

  • STOREILM/STOREM: the new ODT still prints storem d0,d1,(a2) on the STOREILM page and storeilm e10,e11,(a2) on the STOREM page. Each example's result follows the rule of the page it is on, not of the mnemonic it shows. The question to Tommo: which mnemonic did he assemble, and which bytes were written to memory?

Knowledge base files relied on

  • CONTRIBUTING.md section 11 (document intake)

🤖 Generated with Claude Code

Tommo answered the intake questions on 2026-11-03. Converter changes:

- An empty encoding box drawn without a border towards one named field
  now belongs to that field. The original leaves such boxes unmerged to
  center the field name. Fixes Bcc/BRA/BSR (displacement 7-1),
  MOVE2/MOVEH/MOVEX/MOVZ2 (B is 15-12), BFLY/TRANS/UNPACK1632 (d is
  11-9) and DBcc.L/FDBcc.L (displacement 15-1).
- Programming model: PC is 32 bits; CC is the 8-bit condition-code part
  of the 16-bit SR, and SR is not part of the user programming model.
- Index: what gray means in the original (less relevant, removed or
  unused).
- PEOR: PNOT is not an instruction; use peor.w #-1.

The 15 regenerated files drop back to draft, as the converter does for
any changed body.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@WDrijver
WDrijver requested a review from TuKo1982 as a code owner November 3, 2026 15:52
Tommo's updated AC68080PRM.odt (update nov 2026) fixes the confirmed
typos (MOVZ2 example, FDBcc operation, PCMPccB bsel), the "where a0
points to" lines on MOVEH and STOREILM, merges the B-field cells, and
adds a note on MOVEC about grayed-out registers.

Converter: recognize paragraph styles by their properties instead of
their automatic names. LibreOffice renamed the code style P47 to P59 in
this version, which broke the code blocks on the overview pages. Style
attributes without a visible effect (opacity, use-window-font-color) are
left out of the comparison. With the previous ODT the converter still
gives the committed output.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@WDrijver WDrijver changed the title PRM intake: apply the document owner's answers PRM intake: the document owner's answers and the November 2026 update Nov 3, 2026
@WDrijver
WDrijver merged commit 55a0062 into main Nov 3, 2026
2 checks passed
@WDrijver
WDrijver deleted the prm-intake-answers branch November 3, 2026 16:30
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant