Skip to content

CHANGELOG.md

Latest commit

 

History

History
83 lines (52 loc) · 3.49 KB

CHANGELOG.md

File metadata and controls

83 lines (52 loc) · 3.49 KB

Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[Unreleased]

[0.1.5] - 2026-06-01

Changed

  • Release 0.1.5

[0.1.4] - 2026-05-27

Changed

  • Release 0.1.4

[0.1.3] - 2026-05-26

Added

  • Parametric Arrow type strings returned by the information schema are now mapped correctly: timestamp[unit, tz=…] (with scale preserved), duration[unit], decimal128(p, s) / decimal256(p, s) / decimal(p, s), and list<item: T> / large_list<item: T> (including non-nullable item fields). Previously all of these fell back to String.
  • Simple Arrow type aliases extended: halffloat (PyArrow's name for float16), large_string, all four signed integer variants (int8–int64), and all four unsigned integer variants (uint8–uint64) are now resolved correctly.

Changed

  • Default schema for managed databases changed from "public" to "main" to match runtimedb's DEFAULT_SCHEMA_NAME. runtimedb always auto-inserts a main schema into every managed database; using "public" previously created a spurious empty main schema alongside the declared public one.
  • Arrow type construction now goes through PyArrow's type system as the authoritative bridge (pa.DataType → PyArrowType.to_ibis()), replacing manual Ibis type construction.
  • _IN_FLIGHT query-run statuses trimmed to {"running"} — runtimedb QueryRunStatus only emits running, succeeded, and failed; queued and pending are result statuses, not query-run statuses.

Fixed

  • Decimal regex tightened: decimal128? → decimal(?:128|256)? so decimal12(…) is no longer mistakenly matched as a decimal type.
  • Unknown duration unit now falls through to the Postgres parser / String fallback instead of silently defaulting to seconds.

[0.1.2] - 2026-05-24

Added

  • Arrow-style type names (Date32, Float64, Utf8, LargeBinary, etc.) returned by the information schema for Parquet/managed-table columns are now mapped to the correct Ibis types instead of falling back to String.
  • database_id parameter on connect() to bind a pre-existing managed database at connect time, enabling "default" catalog references without calling create_table first.

Changed

  • Managed databases now use the dedicated /v1/databases API instead of the legacy /v1/connections API. create_database and drop_database are unaffected at the call site.
  • create_table now returns a table reference with ("default", schema) as the catalog, matching the SQL catalog prefix required for querying managed tables.
  • create_database no longer silently swallows API errors during the existence check — only a genuine not-found result proceeds to creation.

Fixed

  • Querying managed tables via con.table(..., database=("default", schema)) or con.sql(...) now correctly routes information-schema lookups through the underlying connection id rather than the literal string "default".
  • Cached _database_id is now always overwritten on each create_table / drop_table call, preventing a stale id from being sent in subsequent queries when different managed databases are used on the same connection.

[0.1.1] - 2026-05-19

Added

  • Managed database support via create_managed_database and load_managed_table.

[0.1.0] - 2026-05-06

Added

  • Initial release.