|
| 1 | +[review] |
| 2 | +schema = "sqlquack.recorder-change-review.v1" |
| 3 | +decision = "D284" |
| 4 | +branch = "feat/varchar-binary-cast" |
| 5 | +commit = "7928dad0" |
| 6 | +base = "29f1095d" |
| 7 | +pull_request = 129 |
| 8 | +supersedes = "product-d284-binary-pr129-29f1095d.toml" |
| 9 | +reviewed_at = "2026-11-06T18:32:25Z" |
| 10 | +round = 2 |
| 11 | +verdict = "approved" |
| 12 | +checked = [ |
| 13 | + "delta 29f1095d..7928dad0: 13a2797c (obligations, approved separately), d45bb0e5 and 7928dad0; only casts.rs, typing targets/constants, the CAST rule specs, two test files and the t008 case description change", |
| 14 | + "round-1 blocker fixed: bounded() checks BINARY(n) with binary_length_check, and strict_cast's same-type arm now calls bounded(); only VARCHAR and BINARY targets carry max_length and VARCHAR to VARCHAR is handled earlier in cast_sql, so no other same-type cast changes", |
| 15 | + "built 7928dad0 and converted: CAST(B AS BINARY(2)) and B::BINARY(1) over a BINARY column emit the 100145 check; CAST(B AS BINARY) stays a bare BLOB cast; TRY_CAST(B AS BINARY(2)) refuses; BINARY(67108865) refuses and BINARY(67108864) converts; VARCHAR(3) still emits the 100078 check", |
| 16 | + "the four new given_computed_subscripts rows and the two new analyze rows pass at 7928dad0 (cargo test -p sqlquack --lib, 2 tests, 0 failed)", |
| 17 | + "CAST_SAME_TYPE declares BINARY_TOO_LONG beside STRING_TOO_LONG, and its conditions state the BINARY(n) failure", |
| 18 | + "t008's case description no longer cites a page", |
| 19 | + "B's account of the 28 baseline changes (only data-types-text.binary and ENCRYPT_RAW moved as forms; the 104 refused cast:BINARY statements still refuse) is taken as reported", |
| 20 | + "gate as reported: verify 11,807 cases, 0 failed, 227/227; against main 0 regressions, the same 28 changes; clippy, fmt, fensu 0", |
| 21 | +] |
| 22 | + |
| 23 | +[[review.finding]] |
| 24 | +kind = "note" |
| 25 | +blocking = false |
| 26 | +detail = "BINARY to BINARY(n) rests on the VARCHAR to BINARY(n) receipt until A records cases/binary-to-binary 57820dea (t011 fits, t012 overflow). When it lands it should carry its own error_parity obligation, as noted on 13a2797c." |
| 27 | + |
| 28 | +[[review.finding]] |
| 29 | +kind = "note" |
| 30 | +blocking = false |
| 31 | +detail = "Leaving generic rules (ORDER BY, GROUP BY, COUNT(DISTINCT), UNION ALL, CASE/IFF, COALESCE, DECODE, ANY_VALUE) admitting BINARY on the unsigned byte-order argument is acceptable for this PR; a receipt for ORDER BY and GROUP BY of BINARY would replace the argument with evidence. No change requested." |
| 32 | + |
| 33 | +[[review.finding]] |
| 34 | +kind = "note" |
| 35 | +blocking = false |
| 36 | +detail = "CAST(NULL AS BINARY(n)) now wraps the NULL in a length check that cannot fire; harmless, same as the VARCHAR path." |
| 37 | + |
| 38 | +[review.files_sha256] |
| 39 | +"crates/sqlquack-rules/src/conversion/_helpers/expressions/casts.rs" = "4ee4efda8dddaf556d8c683ae473d5d01b055a59c87ce9ac921a96db7785c7f2" |
| 40 | +"crates/sqlquack-snowflake/src/typing/_helpers/targets.rs" = "41f05cbb7d996c024ee4b1714660b3c2db0ab0c86025d1678f5b528afd626552" |
| 41 | +"crates/sqlquack-snowflake/src/typing/constants.rs" = "9a99fc21abd7f2c69dddbb890b6e08f67e072698e49f5e26e4a14400068662c6" |
| 42 | +"crates/sqlquack-rules/src/catalogue/constants.rs" = "abb52541a3103e5b03ea1eb3db243adc7363c801f9060b9a7774480e45a3f1af" |
| 43 | +"crates/sqlquack/src/analysis/tests/test_coverage_batch.rs" = "03f17a294baf9cb01d2aac06f4ccae19da0468fd2ff7785461a022bb1453737c" |
| 44 | +"cases/translation/functions/cast.cast-as.toml" = "697d0a458b361faa35463b5fe1d37b480d73d644ea7dac1ad0cba80054e5fedc" |
0 commit comments