Skip to content

Upgrade for Datasette 1.0a41 - #64

Draft
asg017 wants to merge 2 commits into
otel/00-lintfrom
otel/01-datasette-1a41
Draft

asg017 wants to merge 2 commits into
otel/00-lintfrom
otel/01-datasette-1a41

Conversation

@asg017

@asg017 asg017 commented Sep 28, 2026 •

Copy link
Copy Markdown
🤖 Claude-generated PR description

Makes the plugin require Datasette 1.0a41 and fixes the test suite against it. The dependency was datasette>=1.0a21 with no upper bound, so fresh installs and CI already resolve to 1.0a41, where 14 of the 20 tests fail.

Stack: lint (#63) → 1.0a41 compatibility (this PR) → supervised background tasks (#65) → empty jobs (#66) → telemetry (#67). The base branch is otel/00-lint.

Changes

  • pyproject.toml: datasette>=1.0a41.
  • CSRF tokens were removed in Datasette 1.0a27. Protection now uses Sec-Fetch-Site/Origin.
    • Removed the five hidden csrftoken inputs from enrichment.html and enrichment_job.html. They still rendered via a legacy shim, but they're dead weight.
    • The tests no longer read the ds_csrftoken cookie or post a token.
  • The row-action menu markup changed. Links now carry role="menuitem" tabindex="-1", so test_row_actions matches the link with a regex that ignores its attributes. It still checks the "1 row selected" page.

Testing

  • pytest: 20 passed, both unpinned and with --with datasette==1.0a41.
  • ruff and black: clean.
  • Manual smoke test against a real server with curl:
    • The enrichment form POST and the pause, resume and cancel POSTs all work without a token.
    • A Sec-Fetch-Site: cross-site POST still gets a 403.

Release note

Datasette 1.0a39 added a default 2s execute_write time limit. It could affect third-party enrichments that do large writes in enrich_batch. There's no code change for it here.

🤖 Generated with Claude Code

asg017 and others added 2 commits September 28, 2026 13:52
Datasette 1.0a27 removed CSRF tokens and 1.0a34+ changed action menu
markup; drop csrftoken inputs and make the row action test robust.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Fixed 100ms sleeps assumed the job had already processed rows, which
flaked on slower CI runners. Also raise the wait_for_job() timeouts
from 1s to 5s for the same reason.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

This branch has not been deployed

No deployments
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