rivet-cli 0.20.0

Rivet: PostgreSQL/MySQL/SQL Server/MongoDB → Parquet/CSV (local, S3, GCS, Azure). Crate name rivet-cli; binary rivet.
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137

# Warehouse-load coverage matrix — the OTHER axis from type-fidelity: not
# "does the type survive source→Parquet" but "does it survive Parquet→WAREHOUSE
# AUTOLOAD". Columns are the four ExportTarget variants (src/types/target.rs), not
# source engines. Cell = { test } | { gap } | { na }.
#
# The bug class this guards (CLAUDE.md "Remediation hints must recover from the
# degraded state" + its native/autoload-truth extension): the resolver must never
# claim a native type the warehouse cannot hold, autoload_type must be the EMPIRICAL
# autoload, and any lossy divergence gets cast_sql only where recovery is real. This
# matrix caught 3 such bugs (Snowflake decimal>38 falsely Ok; DuckDB decimal>38
# mislabeled DECIMAL instead of DOUBLE; ClickHouse decimal>76 falsely Ok) — all fixed.
#
# Coverage depth note: DuckDB/ClickHouse/BigQuery have live round-trip validators for
# PG/MySQL/MSSQL; Snowflake has a live PostgreSQL matrix validator plus a live MySQL
# UInt64-overflow test (the one type PG can't produce — the overflow class is now
# live-proven, not offline-only). Cells marked with an offline resolver/oracle test
# (all_types_oracle.rs, target.rs mod tests) are
# CONTRACT-verified but not always live-round-tripped — a deliberate ratchet target for
# later live Snowflake(MySQL) coverage, tracked in the inventory. The resolver-only
# defensive arms (List<Unsupported>, Interval, negative-scale decimal — unreachable via a
# real Parquet write today) are pinned by offline unit tests so a future arm edit can't
# silently drop them.

engines: [duckdb, bigquery, snowflake, clickhouse]

scenarios:
  - id: standard_scalars
    what: "bool / int / float / date / string / binary — the clean-autoload baseline"
    duckdb:     { test: duckdb_validates_postgres_type_matrix_parquet }
    bigquery:   { test: bq_standard_scalars_ok }
    snowflake:  { test: snowflake_validates_postgres_type_matrix_parquet }
    clickhouse: { test: clickhouse_validates_postgres_type_matrix_parquet }

  - id: uint64
    what: "unsigned 64-bit: overflow class — BQ/SF degrade to a load-schema note, DuckDB/CH hold it natively"
    duckdb:     { test: duckdb_reads_everything_natively }
    bigquery:   { test: bq_uint64_recommends_numeric_warns_overflow }
    snowflake:  { test: snowflake_validates_mysql_uint64_overflow_survives_via_decimal_override }
    clickhouse: { test: clickhouse_holds_uint64_and_naive_timestamp_natively }

  - id: decimal_normal
    what: "decimal(p,s) within every target's precision — native NUMBER/NUMERIC/DECIMAL"
    duckdb:     { test: duckdb_native_type_names }
    bigquery:   { test: bq_decimal_within_numeric_is_numeric }
    snowflake:  { test: snowflake_autoload_degradations_and_native_casts }
    clickhouse: { test: clickhouse_validates_postgres_type_matrix_parquet }

  - id: decimal_over_ceiling
    what: "decimal precision past the target's ceiling — must NOT falsely claim Ok (the bug this matrix caught, 3 engines)"
    duckdb:     { test: duckdb_decimal_over_38_autoloads_as_double_not_a_false_native_decimal }
    bigquery:   { test: bq_decimal_limit_boundaries }
    snowflake:  { test: snowflake_decimal_over_38_fails_not_falsely_ok }
    clickhouse: { test: clickhouse_decimal_over_76_fails }

  - id: time_of_day
    what: "Time64 — no native warehouse TIME type on SF/CH (autoloads as NUMBER/Int64)"
    duckdb:     { test: duckdb_validates_postgres_type_matrix_parquet }
    bigquery:   { test: bigquery_validates_postgres_type_matrix_parquet }
    snowflake:  { test: snowflake_autoload_degradations_and_native_casts }
    clickhouse: { test: clickhouse_time_autoloads_as_int64 }

  - id: timestamp_naive
    what: "naive timestamp — BQ/SF autoload as INT64/NUMBER (recover via cast), DuckDB/CH native"
    duckdb:     { test: duckdb_reads_everything_natively }
    bigquery:   { test: bq_naive_timestamp_is_datetime_native_timestamp_autoload }
    snowflake:  { test: snowflake_autoload_degradations_and_native_casts }
    clickhouse: { test: clickhouse_holds_uint64_and_naive_timestamp_natively }

  - id: timestamp_tz
    what: "tz timestamp — carries the UTC instant; SF needs a session-TZ note"
    duckdb:     { test: duckdb_validates_postgres_type_matrix_parquet }
    bigquery:   { test: bq_tz_timestamp_is_timestamp_ok }
    snowflake:  { test: snowflake_validates_postgres_type_matrix_parquet }
    clickhouse: { test: clickhouse_validates_postgres_type_matrix_parquet }

  - id: timestamp_nanosecond
    what: "nanosecond timestamp (datetime2(7) / timestamp_ns override) — the ns autoload truth per target"
    duckdb:     { test: duckdb_nanosecond_timestamp_is_native_timestamp_ns }
    bigquery:   { test: bq_nanosecond_timestamp_autoloads_as_int64 }
    snowflake:  { test: snowflake_nanosecond_timestamp_recovers_losslessly_at_scale_9 }
    clickhouse: { test: clickhouse_nanosecond_timestamp_is_datetime64_9 }

  - id: json
    what: "JSON — DuckDB/CH hold it, BQ/SF autoload as bytes/string and recover via PARSE_JSON"
    duckdb:     { test: duckdb_reads_everything_natively }
    bigquery:   { test: bq_json_native_is_json_autoload_is_bytes }
    snowflake:  { test: snowflake_autoload_degradations_and_native_casts }
    clickhouse: { test: clickhouse_validates_postgres_type_matrix_parquet }

  - id: uuid
    what: "UUID — autoloads as bytes on BQ/SF/CH, recover via TO_HEX/toUUID; DuckDB native"
    duckdb:     { test: duckdb_reads_everything_natively }
    bigquery:   { test: bq_uuid_resolves_not_fails }
    snowflake:  { test: snowflake_autoload_degradations_and_native_casts }
    clickhouse: { test: clickhouse_validates_postgres_type_matrix_parquet }

  - id: list_scalar
    what: "array of a scalar — REPEATED/ARRAY per target, flatten note where autoload nests"
    duckdb:     { test: duckdb_reads_everything_natively }
    bigquery:   { test: bq_list_is_repeated_native_record_autoload }
    snowflake:  { test: snowflake_autoload_degradations_and_native_casts }
    clickhouse: { test: clickhouse_validates_postgres_type_matrix_parquet }

  - id: enum
    what: "enum labels ride as text (VARCHAR/String) — no divergence"
    duckdb:     { test: duckdb_validates_postgres_type_matrix_parquet }
    bigquery:   { test: bq_standard_scalars_ok }
    snowflake:  { test: snowflake_enum_autoloads_as_text }
    clickhouse: { test: clickhouse_enum_autoloads_as_text }

  - id: unsupported
    what: "a source type with no target mapping is a Fail row, not a panic; the generic remediation contract holds"
    duckdb:     { test: resolver_remediation_contract_holds_across_all_targets }
    bigquery:   { test: bq_unsupported_is_fail_row_not_panic }
    snowflake:  { test: resolver_remediation_contract_holds_across_all_targets }
    clickhouse: { test: resolver_remediation_contract_holds_across_all_targets }

  - id: interval
    what: "interval maps to a text/native type per target (BQ STRING / DuckDB INTERVAL / SF TEXT / CH String), never silently dropped to Fail"
    duckdb:     { test: interval_resolves_to_a_text_or_native_type_per_target }
    bigquery:   { test: interval_resolves_to_a_text_or_native_type_per_target }
    snowflake:  { test: interval_resolves_to_a_text_or_native_type_per_target }
    clickhouse: { test: interval_resolves_to_a_text_or_native_type_per_target }

  - id: list_of_unsupported
    what: "a list whose element type has no mapping must be a Fail row — not a panic, not a dropped column (the nested-fail arm, per target)"
    duckdb:     { test: list_of_unsupported_element_fails_on_every_target }
    bigquery:   { test: list_of_unsupported_element_fails_on_every_target }
    snowflake:  { test: list_of_unsupported_element_fails_on_every_target }
    clickhouse: { test: list_of_unsupported_element_fails_on_every_target }

  - id: decimal_negative_scale
    what: "negative-scale decimal (resolver-only — the Parquet writer rejects it today, but the arm is live code): BQ fails, DuckDB/SF/CH warn + declared schema"
    duckdb:     { test: decimal_negative_scale_is_handled_not_dropped_per_target }
    bigquery:   { test: decimal_negative_scale_is_handled_not_dropped_per_target }
    snowflake:  { test: decimal_negative_scale_is_handled_not_dropped_per_target }
    clickhouse: { test: decimal_negative_scale_is_handled_not_dropped_per_target }