Is this a new component or an extension of an existing one?
Extension of the existing default pipeline node experience (TaskNode / pipeline layout defaults), including how those nodes participate in keyboard focus, selection, and screen reader announcement.
Describe the feature
Keyboard and screen reader users should be able to use default pipeline topology views the same way mouse users can today: move through pipeline steps, understand what each step is and whether it succeeded or failed, and activate a step to select it / open its details.
Concretely, the default pipeline node experience should:
-
Allow users to move focus between pipeline step nodes with the keyboard
-
Show a clear visual indication when a node has keyboard focus (design needed for how focus differs from hover/selected)
-
Announce each focused step’s name and status to assistive technology (for example: that build-image failed)
-
Let users activate a focused step with the keyboard (same outcome as clicking it)
-
Avoid noisy or confusing announcements from non-interactive graph chrome where possible (for example edges currently read as generic “visual elements”)
-
Prefer a predictable focus order through the pipeline where feasible
Jeff Phillips shared an exploration that moves in this direction: https://focus-demo.surge.sh/ (navigate to Pipeline Layout).
Are there visuals for this feature? If applicable, please include examples for each state and for varying widths
Any other information?
Reported against Red Hat OpenShift AI pipeline run topology by GOV UK users on Windows 11 with JAWS / NVDA. Without this, finding a failed step means opening each step’s drawer and reading status manually.
Product tracking:
Discussed with @jeff-phillips-18 — preference is to land this in pf-topology defaults and use this issue to track the work and flush out designs.
Jira Issue: PF-4458
Is this a new component or an extension of an existing one?
Extension of the existing default pipeline node experience (
TaskNode/ pipeline layout defaults), including how those nodes participate in keyboard focus, selection, and screen reader announcement.Describe the feature
Keyboard and screen reader users should be able to use default pipeline topology views the same way mouse users can today: move through pipeline steps, understand what each step is and whether it succeeded or failed, and activate a step to select it / open its details.
Concretely, the default pipeline node experience should:
Allow users to move focus between pipeline step nodes with the keyboard
Show a clear visual indication when a node has keyboard focus (design needed for how focus differs from hover/selected)
Announce each focused step’s name and status to assistive technology (for example: that
build-imagefailed)Let users activate a focused step with the keyboard (same outcome as clicking it)
Avoid noisy or confusing announcements from non-interactive graph chrome where possible (for example edges currently read as generic “visual elements”)
Prefer a predictable focus order through the pipeline where feasible
Jeff Phillips shared an exploration that moves in this direction: https://focus-demo.surge.sh/ (navigate to Pipeline Layout).
Are there visuals for this feature? If applicable, please include examples for each state and for varying widths
Proposed direction / demo: https://focus-demo.surge.sh/ (Pipeline Layout)
Current problematic behavior (RHOAI recording): https://drive.google.com/file/d/17tgaV1*oCIIOVOlRw77zElk*Ne8WGw8c/view
Any other information?
Reported against Red Hat OpenShift AI pipeline run topology by GOV UK users on Windows 11 with JAWS / NVDA. Without this, finding a failed step means opening each step’s drawer and reading status manually.
Product tracking:
Discussed with @jeff-phillips-18 — preference is to land this in pf-topology defaults and use this issue to track the work and flush out designs.
Jira Issue: PF-4458