Skip to content

Nebula railcore overdrive#73

Open
doublemover wants to merge 6 commits intohyperlane-usr-overdrivefrom
nebula-railcore-overdrive
Open

Nebula railcore overdrive#73
doublemover wants to merge 6 commits intohyperlane-usr-overdrivefrom
nebula-railcore-overdrive

Conversation

@doublemover
Copy link
Owner

Summary

G-LEKopXEAEhfUI

General Checklist

  • no mistakes

Copy link

@chatgpt-codex-connector chatgpt-codex-connector bot left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: f07a171175

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".

const sparseUnavailable = Array.isArray(sqliteFtsDiagnostics)
? sqliteFtsDiagnostics.find((entry) => entry?.code === RETRIEVAL_SPARSE_UNAVAILABLE_CODE)
: null;
if (!annEnabled && !vectorOnlyProfile && sparseUnavailable) {

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P1 Badge Stop aborting after successful BM25 sparse fallback

This unconditional sparseUnavailable throw now fires even when sparse retrieval already succeeded via JS BM25 fallback. With SQLite indexes that are missing only chunks_fts (or other sqlite-fts-required tables) but still have token postings tables, the pipeline can produce valid BM25 hits, yet this branch still raises CAPABILITY_MISSING, turning a recoverable FTS unavailability case into a hard failure for ANN-off queries.

Useful? React with 👍 / 👎.

Comment on lines +435 to +439
const selectedModes = PROFILE_MODES.filter((mode) => (
(mode === 'code' && runCode)
|| (mode === 'prose' && runProse)
|| (mode === 'extracted-prose' && runExtractedProseRaw)
|| (mode === 'records' && runRecords)

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P1 Badge Exclude missing optional modes from profile-mix checks

Profile cohort validation is currently driven by requested modes, not by modes that actually loaded an index state, so default/prose searches still include extracted-prose even when that optional index is absent. Because missing state normalizes to default, vector-only code/prose indexes are incorrectly treated as mixed-profile cohorts and rejected unless --allow-unsafe-mix is passed, which blocks valid searches.

Useful? React with 👍 / 👎.

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