Skip to content

chore: release hotdata-langchain v0.15.0 - #96

Merged
rohan-hotdata merged 2 commits into
mainfrom
release/v0.15.0
Sep 1, 2026
Merged

chore: release hotdata-langchain v0.15.0#96
rohan-hotdata merged 2 commits into
mainfrom
release/v0.15.0

Conversation

@rohan-hotdata

@rohan-hotdata rohan-hotdata commented Sep 1, 2026

Copy link
Copy Markdown
Contributor

Summary

Release hotdata-langchain v0.15.0.

Checklist

  • Version bumped in pyproject.toml
  • CHANGELOG.md updated
  • CI green

After merge, run ./scripts/release.sh publish from a clean main checkout.

@rohan-hotdata
rohan-hotdata requested a review from a team as a code owner September 1, 2026 14:37
@rohan-hotdata
rohan-hotdata requested review from anoop-narang and removed request for a team September 1, 2026 14:37
`delete` removing matched rows and inserting nothing is the most consequential
change in this release — the previous wording invited a reading that destroys
rows and reports success. It belongs under Fixed, next to the `keys` validation,
rather than as a footnote. The floor bump now says outright that it tightens what
an existing installation may resolve.
claude[bot]
claude Bot previously approved these changes Sep 1, 2026

@claude claude Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Release metadata is consistent. pyproject.toml, uv.lock and the new ## [0.15.0] heading agree on 0.15.0. The previously unreleased entries now sit under the version heading, and no stale 0.14.0 reference remains in the tree. CHANGELOG.md uses no link reference definitions, so no compare link is needed.

CI checks other than the release metadata check were still queued at review time.

Comment thread CHANGELOG.md
Comment on lines +47 to +55
- **The load tool now says what each keyed mode does to a matched row**, rather than naming
`upsert`, `update` and `delete` together under one clause about matching by key. `delete`
removes matched rows and inserts nothing, and the previous wording left it open to reading as
delete-then-insert by key — a misreading that destroys rows and reports success. Each mode now
states its effect on a matched row and on one that matches nothing.

- **A `keys` entry naming a table the same call is not declaring is refused.** It would
otherwise create a keyless table and report success, and since a key can only be set at
creation, the only fix is to create another database.

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

nit: move both entries into ### Added, or restore the earlier ### Notes placement (not blocking).

Neither entry describes a fix relative to a released version. mode and key on the load tool arrive in this release, per the ### Added entry at line 22, which states the load hardcoded mode="replace" and that the keyed modes were unreachable. keys on the create tool arrives in this release too, per line 14. So no released version carried the ambiguous mode wording, and no released version accepted a keys entry for an undeclared table.

Keep a Changelog reserves Fixed for bug fixes. A reader upgrading from 0.14.0 reads these two entries as regressions in 0.14.0 and finds nothing to have been affected by.

Comment thread CHANGELOG.md
Comment on lines +72 to +73
- **`hotdata-framework>=0.13.0`** (from `>=0.10.0`), which tightens what an existing installation
may resolve. 0.13.0 is where an `append` load became

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

super nit: reflow the paragraph (not blocking). The inserted clause leaves line 73 at 54 characters while the rest of the file fills to the 100-character margin.

Suggested change
- **`hotdata-framework>=0.13.0`** (from `>=0.10.0`), which tightens what an existing installation
may resolve. 0.13.0 is where an `append` load became
- **`hotdata-framework>=0.13.0`** (from `>=0.10.0`), which tightens what an existing
installation may resolve. 0.13.0 is where an `append` load became

@claude claude Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Release metadata is consistent: pyproject.toml and uv.lock both read 0.15.0, and CHANGELOG.md carries a ## [0.15.0] heading under ## [Unreleased]. scripts/extract-changelog.py reads the whole section, so the release notes will include every subsection.

The change since the last review is CHANGELOG text only. Two non-blocking comments are inline.

Only Check release metadata reported a result at review time. The test matrix, lint and floor jobs were still queued or in progress, so this approval makes no claim about them.

@rohan-hotdata
rohan-hotdata merged commit 8d89cfb into main Sep 1, 2026
10 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant