comments

Data license: Public Domain (U.S. Government data) · Data source: Federal Register API & Regulations.gov API

26,299,009 rows sorted by posted_date descending

✎ View and edit SQL

This data as json, CSV (advanced)

These facets timed out: agency_id, submitter_type, posted_year

Next page

CREATE TABLE comments (
    id TEXT PRIMARY KEY,
    agency_id TEXT,
    docket_id TEXT,
    title TEXT,
    submitter_name TEXT,
    submitter_type TEXT,
    document_type TEXT,
    posted_date TEXT,
    posted_year INTEGER,
    posted_month INTEGER,
    last_modified TEXT,
    withdrawn INTEGER,
    object_id TEXT,
    -- S4 (DESIGN_S4 §3; fix 1, 2026-09-13). The design specified this column and its own line 169
    -- said "In a from-scratch build these belong in 05's SCHEMA constant, not in ALTER TABLE" —
    -- the transcription never happened, so import_bulk_comments's FIRST write statement
    -- (UPDATE ... SET family_resolution) raised OperationalError on the real schema while the
    -- selftest passed against a hand-built fixture that had the column. PREREG §1.1 / §1.0.
    -- Values: api_only | bulk_only | bulk_wins  (floor 7.3 requires zero rows left 'unresolved').
    family_resolution TEXT NOT NULL DEFAULT 'unresolved',
    -- ⊞ §128 + §144 SUPPRESSION REASON (item 1, 2026-09-22). WHY A STORED COLUMN when §129's
    -- text_source is deliberately computed at query time: text_source is DERIVABLE from which
    -- table holds the text, so storing it would duplicate a fact the schema already carries.
    -- This is the opposite case — the suppression ERASES its own evidence. After
    -- suppress_withdrawn_text() runs, a hidden row is byte-identical to a row that never had
    -- text, so nothing downstream can reconstruct WHY it is empty. The reader then sees an empty
    -- comment with no explanation, which is exactly what §128 says the suppression must not look
    -- like (queue_inbox 811c065).
    -- Values: NULL = not suppressed · 'withdrawn' (§128) · 'not_posted' (§144).
    -- DISJOINT BY CONSTRUCTION per §144: a row the source flags withdrawn is a §128 row and keeps
    -- that treatment, so 'withdrawn' wins when both causes would apply. Asserted, not assumed.
    -- ⚠ NOT a substitute for comments.withdrawn, which keeps its own meaning ("the source flags
    -- this withdrawn"). A row can be suppressed with withdrawn = 0 — that is the disagreement
    -- case (4dc3169) and is precisely why the served page cannot gate on `withdrawn` alone.
    suppression_reason TEXT
);
CREATE INDEX idx_comments_agency ON comments(agency_id);
CREATE INDEX idx_comments_docket ON comments(docket_id);
CREATE INDEX idx_comments_date ON comments(posted_date);
CREATE INDEX idx_comments_date_desc ON comments(posted_date DESC, id ASC);
CREATE INDEX idx_comments_year ON comments(posted_year);
CREATE INDEX idx_comments_submitter ON comments(submitter_name);
CREATE INDEX idx_comments_subtype ON comments(submitter_type);