Skip to content

RE: verify and test-cover the Understanding stage (item 2) - #118

Merged
dwolfson merged 1 commit into
mainfrom
re/understanding
Sep 17, 2026
Merged

dwolfson merged 1 commit into
mainfrom
re/understanding

Conversation

@dwolfson

Copy link
Copy Markdown
Owner

Summary

  • PLAN-FINISH-REPOS.md item 2 ("Understanding — not really started") was stale: the earlier app.js split had already marked understanding: built: true with a real, honest loadChartsPane() implementation (probes all 8 REPO_CHARTS kinds, distinguishes failed/empty/single-observation/nothing-yet, flags stale data, never fakes a trend from one point).
  • This PR is verification + test coverage, not new feature work — traced a real user question end-to-end through the backend routes, and verified live in a browser (amundsen-io/amundsen: trend chart, single-observation honesty label, staleness flag, chip switching all confirmed working).
  • Added tests/test_next_understanding_pane.py (11 new tests, none existed before), pinning the honesty rules and no-data/error messaging.

Full suite: 4837 passed, 0 failed, 103 skipped.

See docs/design-notes/ITEM-2-UNDERSTANDING-IMPLEMENTED.md for the full account, including what was NOT verified (only one repo checked live, not the full corpus).

Test plan

  • uv run pytest tests/test_next_understanding_pane.py -q — 11/11 pass
  • Full suite: uv run pytest tests/ -q -k "not Postgres" — 4837 passed
  • Live browser check against real Postgres/Egeria

🤖 Generated with Claude Code

…m 2)

The plan doc's "not really started" was already stale: the app.js split
(UNBUILT-STAGES-IMPLEMENTED.md) had already flagged understanding as
built and next/stages/understanding.js already had a complete
loadChartsPane() -- probing all eight REPO_CHARTS kinds, distinguishing
a failed probe / an empty series / a single-observation series / no
chart at all as four different sentences, flagging stale data, and
honestly labelling a one-point series as "first measurement" rather
than drawing a fake trend.

This is verification, not new feature work. Traced a concrete question
("is star growth / commit volume trending?") end to end against the
running app (uv run resource-explorer web, TRELLIS_ANONYMOUS_READ=true
for this session only, shared Postgres/Egeria, amundsen-io/amundsen):
confirmed the stars trend chart draws with an honest not-zero-axis
label, the single-observation top_committers chart shows the "value,
not a trend" caveat, the stale-data flag reads on stars' last point,
and the health radar's cross-model caveat renders live.

Added tests/test_next_understanding_pane.py (none existed before) --
eleven tests pinning the honesty rules, the stale-data flag, the
no-data/error messaging, and that chart selection actually calls
drawChart -- and docs/design-notes/ITEM-2-UNDERSTANDING-IMPLEMENTED.md
recording what was verified, what was found already correct, and what
was scoped out.

Full suite: 4837 passed, 103 skipped, 0 failed.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Signed-off-by: Dan Wolfson <dan.wolfson@pdr-associates.com>
@dwolfson
dwolfson merged commit 55410fc into main Sep 17, 2026
3 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