chore: bump version to 1.3.0-dev#373
Conversation
Follow-up to the v1.2.0 release (#358), per the playbook in RELEASING.md. Generated via: ./scripts/bump-version.sh 1.2.0 1.3.0-dev The script renamed the main install SQL to sql/pg_textsearch--1.3.0-dev.sql, created the stub upgrade file sql/pg_textsearch--1.2.0--1.3.0-dev.sql (library-version check only -- contributors append to this as features land in the dev cycle), and bumped the version reference in pg_textsearch.control, Makefile, src/mod.c, README.md, CLAUDE.md, and the test/benchmark scripts that pin the expected version. The remaining literal '1.2.0' occurrences flagged by the script are intentional historical references (on-disk bm25vector v1->v2 migration notes in src/types/vector.{c,h} and test/sql/vector_v1_rejected.sql, plus the legacy 1.1.0->1.2.0 upgrade SQL filename in Makefile DATA) and are left unchanged. Verified: make builds clean against pg17. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
There was a problem hiding this comment.
LGTM — mechanical version bump generated by the bump-version.sh playbook script.
Extended reasoning...
Overview
This PR bumps the in-development version from 1.2.0 to 1.3.0-dev, generated via the documented ./scripts/bump-version.sh 1.2.0 1.3.0-dev automation per RELEASING.md. It renames the main install SQL file, creates the standard upgrade stub (1.2.0--1.3.0-dev.sql with just the library-loaded check), and updates the version string in the Makefile DATA list, control file, src/mod.c (PG_MODULE_MAGIC_EXT), README/CLAUDE.md, and the test/benchmark scripts that pin the expected version.
Security risks
None. No logic changes, no new code paths — purely string/version updates and a new SQL stub that performs a standard 'library loaded?' check identical to the prior release's pattern.
Level of scrutiny
Minimal scrutiny appropriate. This is a mechanical, scripted version bump that follows an established playbook (RELEASING.md). The diff matches exactly what the bump script produces for this kind of follow-up. The upgrade stub deliberately omits the version-equality check (correctly documented inline: chained upgrades may run as intermediate steps).
Other factors
The bug hunting system found no issues. All literal 1.2.0 references left in the tree are explicitly accounted for in the PR description as intentional historical references (v1→v2 vector migration notes, legacy upgrade SQL filename). The PR is small (11 files), self-contained, and the author verified make builds clean against pg17. Author has correctly sequenced this before the dependent PR #372.
Follow-up to the v1.2.0 release (#358), per the playbook in RELEASING.md. Generated via:
Renames the main install SQL to
sql/pg_textsearch--1.3.0-dev.sql, creates the library-version-check stub atsql/pg_textsearch--1.2.0--1.3.0-dev.sql(contributors append to this as features land), and bumps the version reference inpg_textsearch.control,Makefile,src/mod.c,README.md,CLAUDE.md, and the test/benchmark scripts that pin the expected version.The remaining literal
1.2.0occurrences flagged by the bump script are intentional historical references (on-disk bm25vector v1→v2 migration notes insrc/types/vector.{c,h}andtest/sql/vector_v1_rejected.sql, plus the legacy1.1.0--1.2.0upgrade SQL filename inMakefileDATA) and are left unchanged.Verified locally:
makebuilds clean against pg17.Per
RELEASING.md, this PR should land before the Wikipedia-validation PR (#372) is merged.