Give every evidence entry the same shape - #2531
Merged
dzarlax merged 2 commits intoAug 14, 2026
Merged
Conversation
Opening supporting data from a chart block offered a metric picker; opening the same data from a card, an activity strip or a roster cell did not. Two doors into one dialog, and from the card there was no way to reach a neighbouring metric without closing it and finding another control. A surface now declares the metrics it holds, and every entry opens that set with the metric the reader clicked already selected. The caller's own selection wins over the surface's copy of it, so a card that drills with its breakdown dimensions keeps them. The provider takes the active metric as an option rather than always opening the first target, and falls back to the first when asked for one that is not in the set. Signed-off-by: Alexey Panfilov <alexey.panfilov@constructor.tech>
|
Warning Review limit reached
Next review available in: 53 minutes You've used all free OSS reviews for now. Wait for the free limit to reset to keep reviewing this public repository. How can I continue?After more reviews become available, a review can be triggered using the To avoid repeated limits, reduce automatic review volume by pausing incremental auto-reviews earlier, using label-based review opt-in, excluding WIP or generated PR titles, or requesting reviews manually when the PR is ready. If your team needs uninterrupted high-volume reviews, an organization admin can enable usage-based reviews. How do review limits work?CodeRabbit enforces per-developer PR review limits for each organization. Most developers receive the normal plan review availability. For paid Pro and Pro+ PR reviews, CodeRabbit uses adaptive limits for sustained high-volume activity. When a developer's recent PR review activity reaches the 95th percentile or higher among CodeRabbit users, additional reviews become available more gradually as earlier reviews age out of the rolling window. Please refer docs for additional details. Review details⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Pro Plus Run ID: 📒 Files selected for processing (16)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
dzarlax
enabled auto-merge
August 14, 2026 08:37
ktursunov
approved these changes
Aug 14, 2026
ktursunov
pushed a commit
that referenced
this pull request
Aug 14, 2026
Journey 7 asserted that a cell's dialog offers nothing to switch to. #2531 made every entry open the surface's metric set with the clicked one selected, and for a roster cell that set is the member's own columns — so the assertion described the shape the feature replaced, and the journey failed on it as soon as the dialog was named again and the run got that far. Assert the current contract instead: the picker is there, opened on the column that was clicked. Then switch to another column and check the request still asks about that member — the scope is built from the row, and one built from the wrong axis would answer 200 with somebody else's rows, which is the very thing this journey exists to catch. Signed-off-by: Konstantin Tursunov <Konstantin.Tursunov@constructor.tech>
ktursunov
pushed a commit
that referenced
this pull request
Aug 14, 2026
Journey 7 asserted that a cell's dialog offers nothing to switch to. #2531 made every entry open the surface's metric set with the clicked one selected, and for a roster cell that set is the member's own columns — so the assertion described the shape the feature replaced, and the journey failed on it as soon as the dialog was named again and the run got that far. Assert the current contract instead: the picker is there, opened on the column that was clicked. Then switch to another column and check the request still asks about that member — the scope is built from the row, and one built from the wrong axis would answer 200 with somebody else's rows, which is the very thing this journey exists to catch. Two things the switch has to respect. The dialog is named for the metric on show, so it renames itself and the handle it opened under stops resolving. And a neighbouring column the member has nothing recorded for answers honestly with no records, so the person is what the assertion holds, not rows. The cell's own label is read BEFORE the dialog opens: a modal hides the grid behind it from the accessibility tree, and a role-based locator then resolves to nothing. Signed-off-by: Konstantin Tursunov <Konstantin.Tursunov@constructor.tech>
dzarlax
pushed a commit
to dzarlax/insight-back
that referenced
this pull request
Aug 14, 2026
…ic#2540) * ci(test-stand): build whichever tree the ref changes The stand pins the frontend and the four backend services to their charts' appVersions, which name what main released. A ref that changes one of those trees and still pins it has the lane reporting on code the run never executed. `changes` asked that question for src/backend/ only, so a diff touching just the SPA ran against main's frontend image. dev-compose.sh carries a guard for exactly this case, but it needs origin/main and the stand jobs clone at depth 1, so it returned early and said nothing. Make the two trees symmetric on both sides of the interface. `test-stand up` grows --build-backend and --build-frontend as independent axes, with --build kept as the both-axes alias; `changes` emits one output per tree and the lanes map each onto its own flag. A frontend diff now costs a pnpm build rather than the 26-minute Rust compile --build would have pulled in. The guard now says on stderr when it cannot resolve origin/main, so a shallow CI log no longer reads as if it had vouched for the pinning. Signed-off-by: Konstantin Tursunov <Konstantin.Tursunov@constructor.tech> * fix(frontend): name the evidence dialog for the metric it shows Giving every entry the metric picker (constructorfabric#2531) also gave every entry more than one target, and the multi-target branch names the dialog from a title the caller supplies — which only the timeseries block does. Opening from a card, a chart point or a roster cell therefore fell back to the placeholder "Metric evidence", where it had been the metric's own name. A screen reader now hears the metric again, and the three deployed-stand journeys that locate the dialog by that name pass. The joined title the timeseries block passes still wins where it is given. Signed-off-by: Alexey Panfilov <alexey.panfilov@constructor.tech> (cherry picked from commit d0da982) * docs(test-stand): trim the comments, correct the measured cost The prose around the two build axes said in ten lines what fits in three. The numbers it quoted were also stale. Measured across the last 60 completed runs of this workflow (62 lane jobs, 21 of them on the build path): the bring-up step takes 2.5-4.0 min pinned and 23.8-32.5 min building, and whole jobs take 3-6 min pinned against 25-34 min building. So the compile adds ~28 min rather than the ~26 quoted, pull-mode jobs are nowhere near the 8 min the timeout rationale assumed, and the 75-minute ceiling keeps about 2x over the worst build observed. Signed-off-by: Konstantin Tursunov <Konstantin.Tursunov@constructor.tech> * test(stand): the roster cell dialog offers its row's metrics Journey 7 asserted that a cell's dialog offers nothing to switch to. constructorfabric#2531 made every entry open the surface's metric set with the clicked one selected, and for a roster cell that set is the member's own columns — so the assertion described the shape the feature replaced, and the journey failed on it as soon as the dialog was named again and the run got that far. Assert the current contract instead: the picker is there, opened on the column that was clicked. Then switch to another column and check the request still asks about that member — the scope is built from the row, and one built from the wrong axis would answer 200 with somebody else's rows, which is the very thing this journey exists to catch. Two things the switch has to respect. The dialog is named for the metric on show, so it renames itself and the handle it opened under stops resolving. And a neighbouring column the member has nothing recorded for answers honestly with no records, so the person is what the assertion holds, not rows. The cell's own label is read BEFORE the dialog opens: a modal hides the grid behind it from the accessibility tree, and a role-based locator then resolves to nothing. Signed-off-by: Konstantin Tursunov <Konstantin.Tursunov@constructor.tech> --------- Signed-off-by: Konstantin Tursunov <Konstantin.Tursunov@constructor.tech> Signed-off-by: Alexey Panfilov <alexey.panfilov@constructor.tech> Co-authored-by: Konstantin Tursunov <Konstantin.Tursunov@constructor.tech> Co-authored-by: Alexey Panfilov <alexey.panfilov@constructor.tech>
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
2 participants
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.
Follow-up to #2518.
The problem
The evidence dialog had two shapes. Opening it from a chart block passed every
metric in the block, so the dialog showed a metric picker in its header.
Opening the same dialog from a summary card, a distribution, a peer story, an
activity strip or a roster cell passed one metric, so it showed a plain title.
Same dialog, same table, same export — but it looked like two different
features, and from a card there was no way to reach a neighbouring metric
without closing the dialog and hunting for another control.
What changed
A surface now declares the metrics it holds, through a scope context:
Every entry opens that set with the metric the reader clicked already
selected.
openEvidenceTargetstakes the active metric as an option insteadof always opening the first target, and falls back to the first when handed a
key that is not in the set.
The caller's own selection wins over the scope's copy of it (
withOwnTarget),so a card that drills with its breakdown dimensions, or a chart point that
drills with its bucket period and group filter, keeps exactly what it built —
the scope only supplies the neighbours.
Where there is no scope — a card rendered outside a section — the dialog opens
with one metric, as before.
Verification
pnpm typecheck,eslint,vitest run --project unit(1311 passing)scope, and making the provider ignore the requested active metric, each
fails the test written for it.
picker listing that section's drillable metrics and the card's own metric
selected; switching to another metric reloads the table under the new
heading.
clicked column selected.