I have a long HTML form that I am trying to make accessible to Voiceover. When I click into a field, interact with it, and then click Done on the keyboard, after a short delay Voiceover is focusing on what appears to be the field most centered on the screen rather than the next element. I tried adding explicit tabindex to all the fields and this did not help. It is not always reproducable, it sometimes works as expected.
Is this expected behavior for Voiceover? Would a user not expect "Done" to take them to the next field (seems like maybe a dumb question since there is also a "Next" button, which works fine)? But as a sighted user I know the "Done" button when I am interacting with a field just means I am done with that field and want to go on to the next action, field whatever.
"Done" dismisses editing mode on VoiceOver. A VoiceOver user would not expect that tapping "done" moves the VO cursor.