Repository navigation
LE: model CEEGMTO / CEEUTCO, the offset from GMT (Part of #5003) - #5417
Conversation
…k) (Part of #5003) Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
|
No description provided. |
|
No description provided. |
|
PR #5417 @ 9fb4d71: 3 failed check(s) == db2 1/4 (failure) -- infra == ground-truth (failure) -- real == ground-truth part 1/2 (failure) -- real |
|
CI shepherd: the head of this PR (a6bf4cd) is not the one a maintainer labelled |
|
No description provided. |
|
No description provided. |
Part of #5003 (siblings of #4949, register L6). First unchecked item: CEEUTCO / CEEGMTO.
Change.
CEEGMTO/CEEUTCO(one service, two names) now translate: hours 0, minutes 0, seconds 0.0, success feedback code. The time-zone decision follows L6 assumption (1), which already pins the det clock and the oracle's clock to GMT (+0000), so the offset is zero. Recorded as ASSUMED in L6; no zone table.det/cobolrt/le/Ceegmto.java(both names map to it),gen.pyLIBRARY / LIBRARY_FLOATS (seconds is COMP-2, arg 2), C twin intests/equivalence/le/le_time.c, vectors + Java/C byte-for-byte probe + wiring tests intest_det_le_time.py, L6 page.Class table (LE date/time services).
Evidence. No burned program calls CEEGMTO/CEEUTCO, so the proof is the model table (C twin vs Java, byte-identical) only; IBM argument order from the LE reference.
Gates. test_det_le_time + test_det_crecust_rest: 144 passed;
audit_check.pyall clear;pr_gates.py --ratchets: 8 pass, 0 fail, 0 skipped. No golden bless (no scanned parsing logic changed).Sibling gaps. CEESECI (needs 9-arg COMP-2 path), CEEDYWK, CEESCEN (shared century window state read by CEEQCEN/CEESECS
YY), CEEFMDA/T/V + null pictures (country-default table, IBM leaves contents open), CEE2526/2527 (C model cannot see output size). None has a burned consumer.🤖 Generated with Claude Code