What's happening
examples/fiori/purchase-info-records-report.flow.yaml (header comment
around line 13, step around lines 48-55) says: "Two steps... left as flagged
TODOs... not guessed" — explicitly contrasting this flow with a guess, because
the source UAT PDF's own tester hit the same ambiguity live.
Only one of the two steps actually is an unresolved TODO. The other resolves
to a concrete selector, Click "Display Purchasing Info Record", with its own
inline comment: "This is a best-effort guess at the tile's visible text, not
a confirmed label" — which is a guess, directly contradicting the file's own
header claim.
If we don't fix it
The header comment is false, and it's false about the specific thing this
example exists to demonstrate (deriving flows from source material without
guessing what the live app hasn't confirmed). Anyone reading the header for
"which steps are safe to trust as-is" gets the wrong answer for this step.
If we fix it
Either the comment stops claiming "not guessed" for a step that is guessed, or
the guessed selector is replaced with an actual TODO like its sibling — one
line, low risk either way.
Before closing
Whichever fix is chosen, the header comment and the step's own inline comment
must agree with each other.
What's happening
examples/fiori/purchase-info-records-report.flow.yaml(header commentaround line 13, step around lines 48-55) says: "Two steps... left as flagged
TODOs... not guessed" — explicitly contrasting this flow with a guess, because
the source UAT PDF's own tester hit the same ambiguity live.
Only one of the two steps actually is an unresolved TODO. The other resolves
to a concrete selector,
Click "Display Purchasing Info Record", with its owninline comment: "This is a best-effort guess at the tile's visible text, not
a confirmed label" — which is a guess, directly contradicting the file's own
header claim.
If we don't fix it
The header comment is false, and it's false about the specific thing this
example exists to demonstrate (deriving flows from source material without
guessing what the live app hasn't confirmed). Anyone reading the header for
"which steps are safe to trust as-is" gets the wrong answer for this step.
If we fix it
Either the comment stops claiming "not guessed" for a step that is guessed, or
the guessed selector is replaced with an actual TODO like its sibling — one
line, low risk either way.
Before closing
Whichever fix is chosen, the header comment and the step's own inline comment
must agree with each other.