Skip to content

docs(curators): bring both documents up to today's work - #240

Merged
adamjohnwright merged 2 commits into
mainfrom
docs/curator-documents-for-today
Sep 18, 2026
Merged

adamjohnwright merged 2 commits into
mainfrom
docs/curator-documents-for-today

Conversation

@adamjohnwright

Copy link
Copy Markdown
Contributor

Both documents were last touched on 16 September and today changed enough to make
parts of them wrong.

Two corrections

The label. CURATOR-REPORT.md described "the Species and Interactors
controls". That label went back to Overlay today (#228) — it and its position
were decided at EBI — so the document named a control that no longer exists under
that name.

A stale figure. It claimed 635 shapes come out of R-HSA-109606's
PowerPoint export. Re-measured from ppt/slides/slide1.xml today: 974, with
zero <p:pic> and no embedded media. The claim the curators were actually testing
— shapes rather than one picture — holds; the number did not (#167).

Added, all things a curator will meet

Testing document

Two rows for today's fixes, including why the resource count is covered by a
unit test: a browser test of it passed with and without the fix, which is
worse than no test.

🤖 Generated with Claude Code

adamjohnwright and others added 2 commits September 18, 2026 03:17
`CURATOR-REPORT.md` said the keyboard-reachable controls were "Species and
Interactors". That label went back to **Overlay** today (#228) -- it and its
position were decided at EBI -- so the document was describing a control that no
longer exists under that name.

It also claimed 635 shapes come out of R-HSA-109606's PowerPoint export.
Re-measured today: **974**, counted from `ppt/slides/slide1.xml`, with zero
pictures and no embedded media. The claim the curators were testing -- that it is
shapes rather than one picture -- holds; the figure was stale (#167).

Added, all of them things a curator will meet:

- the count beside each resource now follows the pathway on screen, and why the
  old numbers were plausible rather than obviously wrong
- a selection survives opening a pathway from inside the diagram; the flag was
  never the problem
- illustration downloads work again, **with the caveat that the fix is a cache on
  this machine** -- so an illustration that still fails is a gap in what is
  cached, not a fault in the site (#230)
- Perturb offers something: its catalogue was never wired up here, so the button
  appeared and did nothing. Eleven pathways now
- if you leave the email box empty when asking for analysis reports, the reports
  are still made and no mail is sent, with nothing saying so (#238, #168)

`RELEASE-TESTING.md` gains the two rows today's fixes are covered by, including
why the count is a unit test: a browser test of it passed with and without the
fix, which is worse than no test.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
I wrote that Perturb now offers something, having wired up DeltaSignal's
catalogue on this box. Checked it against beta rather than against the box:

    buttons offered on beta.reactome.org/PathwayBrowser/R-HSA-69620:
      ["query_stats Analyze"]

No Perturb. `deltaSignal: true` is set on the **development** profile only, and
the comment beside it says why -- `proxy.conf.js` routes /api to a dev-server
proxy that does not exist in a built artifact, so enabling it elsewhere would
offer a button whose every call fails. Beta serves a production-shaped build.

So the catalogue work is real and the endpoint answers, but a curator will not see
any of it. A paragraph telling them to look would have sent them hunting for a
button that is not there.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
@adamjohnwright
adamjohnwright merged commit 770b735 into main Sep 18, 2026
6 checks passed
@adamjohnwright
adamjohnwright deleted the docs/curator-documents-for-today branch September 18, 2026 03:31
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