The register of 24 September, carried forward to v0.2.5. Every entry of the v0.2.4 register keeps its ID and is given its status on the new candidate; findings first recorded against v0.2.5 are added under new IDs. Each entry keeps the source’s own severity and cites where it came from; nothing is re-graded here.
Candidate: Gateway v0.2.5, c99ff8ce86a7c5d2ad343d62232c843f627c9d0c
Previous register:
RFP-003 aggregate findings on v0.2.4 (2026-09-24) (v0.2.4, 98f492ad). Entries that did not change are summarised here; their full text is in that register.
| Key | Source document | What it is |
|---|---|---|
| A | ‣ | Coverage grading of the twelve review-map test cases. Not re-issued for v0.2.5. |
| B | ‣ | Requirements under M1, M3 and M6 that lack evidence, with proposed test cases. Not re-issued for v0.2.5. |
| C | ‣ | Local run of the five release-runnable BTC cases on v0.2.4 |
| D | ‣ | Conformance hunt on v0.2.4, findings /1 to /15 |
| E | ‣ | Local run of the same five BTC cases on the v0.2.5 bundle |
| F | ‣ | Conformance audit of v0.2.5, findings #1 to #16, two independent passes plus merge adjudication |
| G | ‣ | A manual swap on the v0.2.5 bundle that never locked LEZ, findings 3.1 to 3.5 |
| H | ‣ | Comparison of D and F against the source diff, with corrections to F |
| I | issue #103 | Logos review notes on v0.2.5 of 30 September, items 1 to 15, filed as non-blocking for M1 and M3 |
| GW | ../evidence/regression-v0.2.5-20260928/ |
Gateway’s own regression record for the release (RESULTS.txt, PROVENANCE.md). Cited as Gateway-reported, not as Logos evidence. |
RFP baseline: logos-co/rfp@4e9b3c3d.
Review map: Logos review map gist,
last updated 22 September; it has no v0.2.5 results.
CODE code defect
with a failing test, PKG packaging, deployment or bundle-script defect,
APP released app defect, COV test coverage gap, DOC missing document
or submission artifact, SPEC spec wording drift, DISP finding disputed
between the two conformance passes, DEC decision Logos has to take. New
in this register: CAP capability the RFP requires that the shipped
product does not reach, NODE Node behaviour defect seen in a run, with no
failing test.—. Where two sources grade the same
defect differently, both grades are shown.98f492ad..c99ff8ce for this register, where no v0.2.5 document
re-examined the entry. Re-run means the v0.2.4 failing test was run
again on the v0.2.5 tree on 2026-10-02 (section 13).| Class | v0.2.4 | Closed | Partly closed in v0.2.5 | New | v0.2.5 total |
|---|---|---|---|---|---|
| CODE code defects | 4 | 1 | 1 | 2 | 6 |
| PKG packaging and bundle | 5 | 0 | 1 | 3 | 8 |
| CAP missing capabilities | — | — | — | 3 | 3 |
| NODE Node behaviour | — | — | — | 4 | 4 |
| APP released app defects | 6 | 0 | 1 | 0 | 6 |
| COV test coverage gaps | 17 | 2 | 3 | 1 | 18 |
| DOC documents and artifacts | 9 | 1 | 2 | 1 | 10 |
| SPEC spec drift | 2 | 1 superseded | 0 | 0 | 2 |
| DISP disputed | 3 | 0 | 0 | 4 | 7 |
| DEC decisions for Logos | 4 | 1 | 0 | 1 | 5 |
| Total | 50 | 5 closed, 1 superseded | 8 | 19 | 69 |
Of the 69 entries, 63 are not closed. Fifteen of the 19 new entries (those from F and G, and PKG-8) concern code and scripts that are the same in v0.2.4, so they are defects found later, not defects v0.2.5 introduced. Of the other four, NODE-4 is a consequence of the #97 fix, COV-18 and DOC-10 concern tests and records that v0.2.5 added, and DEC-5 is upstream.
Closed in v0.2.5:
The five release-runnable BTC cases again pass in both directions on the
v0.2.5 bundle (E). The ci workflow is green on the tagged commit (DOC-8).