After the fix for bug 170981 it becomes impossible for a page to tell when the visual viewport changes (eg. user has pinch-zoomed, OSK is displayed). There are a few rare use cases where this is really desirable. For example when composing a tweet on twitter.com in Chrome the toolbar (add picture, location, characters left, etc.) sticks just above the keyboard. On Safari Twitter uses a worse UI where the add picture button is just at the bottom of the text, and the characters left is at the top right.
To address these use cases we propose the VisualViewport API: https://github.com/wicg/viewportapi
What do WebKit folks think about the use case here, potentially worth solving?
Regardless of whether the use case seems worth solving at the moment, does the VisualViewport API seem like a reasonable design?
The API seems reasonable to me.
Maybe these developments should be announced on WebKit dev? I only see two messages about visual viewports:
(In reply to Frédéric Wang (:fredw) from comment #3)
> Maybe these developments should be announced on WebKit dev? I only see two
> messages about visual viewports:
Good idea, I sent a message to webkit-dev:
Please make sure to have an experimental feature flag, ideally runtime rather than compile time. (Maybe you already did this, I only briefly skimmed the patches in the subtask bugs.)
(In reply to Maciej Stachowiak from comment #5)
> Please make sure to have an experimental feature flag, ideally runtime
> rather than compile time. (Maybe you already did this, I only briefly
> skimmed the patches in the subtask bugs.)
Yes, the patch in bug 179385 adds a runtime experimental feature flag for this.
The feature has been implemented. The remaining test failures are tracked by https://bugs.webkit.org/show_bug.cgi?id=179794, and the change to make zooming invisible is tracked by https://bugs.webkit.org/show_bug.cgi?id=170981