Rebuild a display file whose program writes a format without output fields #602

Manually merged
binjovi-bot merged 2 commits from rpg-9-d4-outfmt-20261022 into trunk 2026-10-22 13:18:12 +00:00
Owner

Round 9 track D, D4 follow-up: the exam's real blocker. A display file whose program writes a format that has no output fields.

The held-out exam reported 1 dds blocking TODO that was not the *SRCSTMT refusal (#600 changed nothing in the exam). #601's fixed kinds named the family, then one more split named the refusal: display_output_formats, "the output records are not the formats that the program writes". Public reproduction: d4-out-const (CR1 has an output field, CR2 only a constant; the program writes both with EXFMT). The output block holds one record (the plan sees 2 lines for CR1); CR2 has no output specification, and the two EXFMT calls differ only in symbol ids, so the object does not say which written format owns the record.

Rule: the records go to the first written formats and the member carries TODO [placeholder] the formats of the output records (it compiles; the round trip tests the choice). More records than written formats stays blocking. The commit pair is the probes (d4-out-const, d4-out-input) and the fix. d4-out-input (a second format with only an input field) is a second gap, the place of the input lines is not proved, held in public for the next PR.

Gates against the baseline of trunk before this PR (the D4-fix binary):

gate result
full Build v22 check exit 0
twins 393 -> 395 groups (the two probes), 0 missing, 0 regressions
all-groups 0 missing, 0 regressions, 0 worse
emission worse [], newly clean: d4-out-const; failures 701 -> 703 (the two failures of d4-out-input)

Exam with this binary (counts only): the display member is rebuilt (dds_files 1, blocked 0, blocking TODOs 0); the exam still fails on the host with 71 failure comments. The largest families: 29 "statement names a field of a file", 22 unresolved, 9 block part of a failed block.

Round 9 track D, D4 follow-up: the exam's real blocker. A display file whose program writes a format that has no output fields. The held-out exam reported `1 dds blocking TODO` that was not the `*SRCSTMT` refusal (#600 changed nothing in the exam). #601's fixed kinds named the family, then one more split named the refusal: `display_output_formats`, "the output records are not the formats that the program writes". Public reproduction: `d4-out-const` (CR1 has an output field, CR2 only a constant; the program writes both with EXFMT). The output block holds one record (the plan sees 2 lines for CR1); CR2 has no output specification, and the two EXFMT calls differ only in symbol ids, so the object does not say which written format owns the record. Rule: the records go to the first written formats and the member carries `TODO [placeholder] the formats of the output records` (it compiles; the round trip tests the choice). More records than written formats stays blocking. The commit pair is the probes (`d4-out-const`, `d4-out-input`) and the fix. `d4-out-input` (a second format with only an input field) is a second gap, `the place of the input lines is not proved`, held in public for the next PR. Gates against the baseline of trunk before this PR (the D4-fix binary): | gate | result | |---|---| | full Build v22 check | exit 0 | | twins | 393 -> 395 groups (the two probes), 0 missing, 0 regressions | | all-groups | 0 missing, 0 regressions, 0 worse | | emission | worse [], newly clean: d4-out-const; failures 701 -> 703 (the two failures of d4-out-input) | Exam with this binary (counts only): the display member is rebuilt (`dds_files` 1, blocked 0, blocking TODOs 0); the exam still fails on the host with 71 failure comments. The largest families: 29 "statement names a field of a file", 22 unresolved, 9 block part of a failed block.
binjovi-bot manually merged commit 5e86887676 into trunk 2026-10-22 13:18:12 +00:00
Sign in to join this conversation.
No reviewers
No labels
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference
sean/sourcerer!602
No description provided.