Release: coordination of recent A11y and element tree changes/proposal

[mdjastrzebski]

Recently we have a lot of activity related to accessibility as well as composite/host elements handling. All of these propositions seem to move RNTL into right direction due to more expressive queries (*ByRole), better matching accessibility behaviour of RN/iOS/Android, and increasing compatibility with RTL/DTL. However, the number of changes is getting high, some of which might be breaking changes as well, so we should do some planning in terms of minor/major releases.

My initial proposal is as follows:

Next minor or patch releases (v11.x.y) - non breaking changes

• Query by role and accessibilityState and improve ByRole query errors #1161 (issue Supports accessibilityStates in ByRole queries #1134 ) support states in *ByRole queries
• feat: option for queries to respect accessibility #1064 (issue Provide query option to respect accessibility #970) with hidden: true, i.e. having same default behaviour as current queries (resolves Provide query option to respect accessibility #970)
• fix: *ByA11yState default value false value for busy, disabled & selected state #1166 (issue bug: *ByA11yState query ignores default state values of false #1165) bug: *ByA11yState query ignores default state values of false

Next major release (v12.0.0)

• feat: option for queries to respect accessibility #1064 switching hidden: false to match default DTL/RTL default
• Effect of accessible prop on *ByRole + hidden, inInaccessible, etc #1163 ignore accessible={true} children when hidden: false, potentially fixes Querying by role with name on nested touchables return multiple elements #1152
• feature: make all (safe) queries return host components #1132 *ByText and *ByDisplayValue, *ByPlaceholder to return host components instead of exported composite Text, TextInput
• byRole queries don't find element that don't have an explicitly set accessibilityRole prop #1130 default roles for some elements based on observed RN/iOS/Android behaviour

Other PRs in the pipeline are not related to a11y/element tree and do not introduce breaking changes, so we might merge them at any moment, perhaps for minor release if they are production ready. I did not include them here to reduce complexity of the discussion.

@thymikee @AugustinLF @pierrezimmermannbam does that above plan for upcoming releases seems reasonable for you, would you want to include some more features/fixes or exclude anything mentioned above. Looking forward for your feedback.

[AugustinLF]

We've also had two valuable PR things merged since last release:

• configure() in feature: configure API from DTL #1141
• fix to text in QueryByRole feature: configure API from DTL #1141

Overall, I'm a proponent of releasing minor/patches as often as needed (as long as the release process is not too much of a burden). So I don't feel very strongly on bundling PRs together. For instance, if we get #1130 done before any breaking change, I'd suggest to release it!

[mdjastrzebski]

@AugustinLF I agree on releasing minor/patches more frequently, so I've updated the original description to make that clear. What was my intention is to band breaking changes with common a11y/element tree theme into one single major release, instead of triggering a higher number of major releases.

[mdjastrzebski]

Closing as no longer relevant