GxPlex

us-21-cfr-part-820-2026-09-26

pending review ATTRIBUTE RESIDUE (windowed) — all 14 remaining span(s) sit inside attribute text the corrected tag matcher removes; raw body not retained, so this is a windowed read and not a re-normalisation

Confirmed, derived: the revision question, on 2026-10-11, by rule candidate_states sha256:4013717a8155baa8. Confirmed, derived: the revision question, on 2026-10-20, by rule candidate_states sha256:4994a34e2f93baa7. Confirmed, derived: the revision question, on 2026-10-24, by rule candidate_states sha256:3fc3633b45e282e6. Confirmed, derived: the revision question, on 2026-11-02, by rule candidate_states sha256:b6e855c8950bb7f5.

What the confirm gate read

The gate was not run. A rule resolved this candidate, and a confirm and a decline are contradictory verdicts about one candidate, so no confirm is computed where a rule has already answered.

What was decided

ATTRIBUTE RESIDUE (windowed) — all 14 remaining span(s) sit inside attribute text the corrected tag matcher removes; raw body not retained, so this is a windowed read and not a re-normalisation

Decided by nobody yet .

What the bytes were of

Source
21 CFR Part 820 (QMSR)
Hashed from
https://www.ecfr.gov/current/title-21/chapter-I/subchapter-H/part-820
The row declares
https://www.ecfr.gov/current/title-21/part-820 The hash is not of the declared address. A wrong referent produces the same difference as a revision.

Evidence

Observed
2026-09-25 → 2026-09-26
Hash before
94a77d5625e80437cff61773776997c68ef0e942838a5c000c812e9dfc315ce1
Hash after
ca64082d7f232821febd374e5e6c9805ec039ac27389635d71d1d8580d9538fd
Bytes differing
2050
Detection class
status change
Retained text
both snapshots retained

Mechanical change class

The class of the difference between the two retained texts. It describes what moved. It does not appraise the change, and there is no ordering between the four classes; a candidate may hold several. It is a different field from the detection class above, which is read from the observation records rather than from the two texts.

No class could be derived: all 1 differing span are markup attribute residue retained by a superseded tag matcher, not document text.

Derived over 1 differing span: 1 excluded as markup attribute residue, 0 eligible. The classification reports classes and counts and never the changed text itself: reproducing a publisher’s words is quotation, and this page is not a quoting surface.

A candidate is a pair of digests that differ, and nothing more. This page reports what differed, where the bytes came from, and which rule was applied. It does not say what any change means.

Last updated: