Grow e2e suite with 6 more user stories (8 -> 14 tests) - #88
Merged
Conversation
Presented 7 candidate live-coverage gaps (each a real feature shipped in a recent PR, previously proven only by a mocked-fetch unit test or a one-off manual Docker/Podman screenshot); picked all but the separate-host/CORS one, which is a different scale of effort (needs genuine container builds or two CORS-enabled servers, not this single-Flask-process harness) and deserves its own future pass. - Delete a mask via its hashtag chip's own remove control, not just the on-image rect's delete button -- never driven live before. - Single view's Baseline tab is genuinely `disabled` with no baseline yet, mirroring the existing Approve-checkmark disabled-proof pattern -- jsdom can't render real interactivity, only a live browser can prove a DOM `disabled` property. - Images actually fill their pane's width rather than shrinking to their 480px natural capture size -- a regression guard for the exact CSS bug (InteractiveImagePane's inline-block -> w-full fix) the redesign shipped and fixed, previously confirmed only by eye. Dual view's own per-pane width is capped below 480px by the page's intentional max-w-5xl regardless of browser width, so the strongest direct "wider than natural size" proof lives in Single view instead -- an assumption verified empirically (the first draft asserted >480px in Dual view too and correctly failed at a real, expected 446px, catching my own wrong assumption before it shipped). - A mask category's deletion is refused while a snapshot is still tagged with it. - Bulk approve reports a genuine partial failure: one candidate image's file is deleted directly from the shared data dir between processing and approving, so the backend's approve_snapshot genuinely 409s for that one snapshot -- not a simulated failure. - A full capture/approve/browse flow against a second, fully self-contained Flask process spawned with PPS_API_TOKEN set (own data dir, own port, own viewer-proxy server, torn down in a local finally block) -- confirms both that an unauthenticated request is genuinely rejected and that the viewer's Settings auth-token field unlocks the real flow live. Auth is deliberately never turned on on the shared backend the other 13 tests reuse, since that would break every one of their existing unauthenticated calls. Verification: all 14 tests run stably across 3 consecutive full runs; the three genuinely new techniques (the width-regression guard, the real-file-deletion failure trigger, and the Baseline-disabled proof) were each mutation-tested against the real product code -- reverted the relevant fix, confirmed the new test fails for the right reason, restored, confirmed green again. Full root-level lint/test/e2e/backend suite green (3 client + 55 viewer + 14/14 e2e + 142 backend). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01NrQeh9CXrigN1cvpJuyzy4
stolau
pushed a commit
that referenced
this pull request
Jul 27, 2026
…flict Three PRs landed on main after this branch forked (#86 self-hosting, #87 mask popup, #88 six more e2e stories), two of which touched the same examples/demo-app/CODEMAP.md paragraph this branch also edited. index.html and style.css merged cleanly with no conflict; CODEMAP.md's prose needed manual reconciliation -- kept main's up-to-date "fourteen tests" description and spliced this branch's .thumbs gallery sentence back into the site/ bullet it belongs to. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01NrQeh9CXrigN1cvpJuyzy4
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Follow-up on the previous e2e-growth PRs. Surveyed 7 candidate user stories for real,
currently-unproven-live gaps -- each a feature that shipped in a recent PR but was only ever
verified by a mocked-fetch unit test or a one-off manual Docker/Podman screenshot -- and
implemented all but the separate-host/CORS one, which is a different scale of effort (needs real
container builds or two CORS-enabled servers, not this single-Flask-process harness) and deserves
its own future pass rather than being bundled in here.
button -- never driven live before.
disabledwith no baseline yet, mirroring theexisting Approve-checkmark disabled-proof pattern.
(
InteractiveImagePane'sinline-block->w-full) the redesign shipped and fixed. Worthcalling out: my first draft asserted ">480px" in Dual view too and it correctly failed at a
real, expected 446px (Dual view's per-pane width is capped below 480px by the page's own
intentional
max-w-5xl, regardless of browser width) -- catching my own wrong assumption beforeit shipped, not a product bug.
directly from the shared data dir between processing and approving, so the backend's
approve_snapshotgenuinely 409s for that one snapshot, not a simulated failure.PPS_API_TOKENset (owndata dir, own port, own viewer-proxy server, torn down locally) -- confirms both that an
unauthenticated request is genuinely rejected and that the viewer's Settings auth field unlocks
the real flow live. Auth is deliberately never turned on on the shared backend the other 13
tests reuse.
Test plan
trigger, Baseline-disabled proof) each mutation-tested against the real product code --
reverted the relevant fix, confirmed the new test fails for the right reason, restored
npm run lint --workspacescleannpm test --workspaces-- 3 client + 55 viewer, unaffectedruff check+pytest-- 142/142, unaffectedexamples/demo-app/CODEMAP.mdupdatedCo-Authored-By: Claude Sonnet 5 noreply@anthropic.com
Claude-Session: https://claude.ai/code/session_01NrQeh9CXrigN1cvpJuyzy4