Repository navigation
D-VAS R2: mailbox selections (A3), versioned coordinate over MapTable (B2); B1 blocked - #2062
Conversation
…concat); tests before the run Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01XAqKMMWgnUU9g7Ldyn6sbk
Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01XAqKMMWgnUU9g7Ldyn6sbk
|
Important Review skippedAuto reviews are disabled on base/target branches other than the default branch. Please check the settings in the CodeRabbit UI or the
|
Codex Review Summary: ✅ CompletedThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01XAqKMMWgnUU9g7Ldyn6sbk
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: a8e73ac813
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
…r row Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01XAqKMMWgnUU9g7Ldyn6sbk
…t (review finding), before the fix Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01XAqKMMWgnUU9g7Ldyn6sbk
… the socket's own space Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01XAqKMMWgnUU9g7Ldyn6sbk
…and u64 lanes; before the run Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01XAqKMMWgnUU9g7Ldyn6sbk
…ls named Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01XAqKMMWgnUU9g7Ldyn6sbk
Stacked on #2061 (which is stacked on #2060).
D-VAS-A3,D-VAS-B2;D-VAS-B1is BLOCKED with its source finding. Ledger:.claude/plans/d-vas-implementation-v1.md, section R2. Draft; not for merge without a merge decision. A4 (typed selection folds) is not started.Head
eec7280cc. Cratelance-graph-mailbox-lanes: 12 tests, debug and release, 0 failed; clippy-D warningsclean.A3: the socket issues its own selections
MailboxLanesimplements DT64'sSelectionSource. It issues position lists itself, from the published columns:issue_order(column): the populated rows sorted by a column, ties in row orderissue_where(column, Keep): a filter by one unsigned column comparison, in row orderissue_concat(a, b): one issued selection followed by another, the only way a bag with repeats arisesA caller cannot hand in positions. The lists are views of one publication: they live in the socket and end with the borrow. The mailbox itself has no home for an order and none was added. The 65,537th name is refused, never aliased.
dr256_ternlog_lab::OrderIndex/SealedOrderare not reused: lab crate,i32keys only.tests/selection.rs, 3 tests on a realMailboxSoA<64>(48 rows). Oracle: a sort of(value, row)pairs, a filter and per-row occurrence counts overowner.meta. Six disable runs, all exit 101: sort removed;<as<=; concatenation deduplicating; the frame check removed; the name wrapping; the row-order/scope check removed.B2: one id, unchanged bytes, two coordinate versions
coords::Coordinatespairs one tile of DT64's existingMapTableat one sealed version with a lane binding. No table, id storage or version counter is added. The source version (the publication) and the coordinate version (the map) are separate: advancing the map writes nothing to the mailbox and leaves the lane binding valid.tests/coords.rs, 2 tests: 48 ids read through assignment 1, the map advances, the same ids read through assignment 2 with the source bytes equal before and after; the first binding is thenStaleMappingfor all 48 ids while its assignment stays readable as history. Refused: an unmapped id, a row number offered as an id, another table's frame, an unknown tile, a row past the lane, a repeated id. Three disable runs, all exit 101: the currency check removed; the row bound removed; the current frame taken at read time.Limit: the id is the map's own
u64, not aNodeGuid, and the test fills the map. No authority fills it for a mailbox.B1:
NodeGuidto a row is BLOCKEDMailboxSoaView::row_for_local_key(u64). Its doc names the authority: an owner that stores keys, "the in-RAMMailboxSoA, once it carries alocal_keycolumn".MailboxSoAhas no such column and does not override the defaultNone.MailboxSoA<N>: a layout change, excluded here without a separate decision.NodeGuid::local_key(), the V1family ++ identitytail; it discriminates only after the 10-byte prefix is resolved.MapTablekeys au64inside au32tile, and no authority maps a GUID prefix to a tile.Nothing was built in its place.
Not established
A selection or coordinate binding that outlives its publication (see #2061); filters beyond one unsigned comparison; typed folds over
u32/u64lanes; cost.🤖 Generated with Claude Code
https://claude.ai/code/session_01XAqKMMWgnUU9g7Ldyn6sbk