-
Notifications
You must be signed in to change notification settings - Fork 0
Record the complete VoiceOver spoken-output regression #2
Copy link
Copy link
Open
Labels
accessibilityAccessibility barriers, support, and inclusive designAccessibility barriers, support, and inclusive designhelp wantedExtra attention is neededExtra attention is neededneeds-at-verificationRequires manual or independent assistive-technology verificationRequires manual or independent assistive-technology verificationpriority:P0Blocks independent completion of a core task or release gateBlocks independent completion of a core task or release gate
Description
Activity
Metadata
Metadata
Assignees
Labels
accessibilityAccessibility barriers, support, and inclusive designAccessibility barriers, support, and inclusive designhelp wantedExtra attention is neededExtra attention is neededneeds-at-verificationRequires manual or independent assistive-technology verificationRequires manual or independent assistive-technology verificationpriority:P0Blocks independent completion of a core task or release gateBlocks independent completion of a core task or release gate
Exact campaign target
Validate the DSH core Web tasks with a person directly operating and listening to VoiceOver on physical macOS.
Do not begin until the campaign manifest says
open. The launcher must report:0.1.2-alpha.2:5803bfcfdd502adac26ae9b8eec12d6aed263ec60.1.0-beta.6:6aed71615edd1db1ec5b12897e1ad40b79294c78dsh-core-at-lab/1.0.0-draftThe previous candidate text in this Issue is superseded. Historical comments remain valid only for the versions they name.
Primary environment
Use AT-CORE-LAB.md. Record these nine claim-eligible task IDs without renaming them:
discover-structurenavigate-sessionssearch-sessionsadjust-layoutswitch-session-viewread-conversationinspect-trajectoryconfigure-settingsedit-composer-draftRequired observation
For every task, retain pass, partial, or fail; independent/effective/safe completion; actual VoiceOver speech and keyboard observations; rotor or Quick Nav behavior when used; focus and VoiceOver cursor before and after important transitions; assistance; workaround; and the smallest reproducible barrier. A modality may be claimed only when directly observed on every claimed task. Do not infer output from captions, DOM, accessibility trees, automation, screenshots, or AI operation.
Use only the disposable synthetic Sessions. Never publish the one-use sign-in URL, raw speech history, recordings, logs, prompts, credentials, personal paths, or participant data.
Acceptance boundary
claim: noneand are not rerun or rewritten to hide a barrier.a11y-at-testedrecord must pass the public human-evidence validator and public review for this exact cohort.