Reported by mdcurran on 2015-01-14 05:10
Currently for MSHTML live regions with relevance of text have all text nodes within the deepest element spoken when it changes. We should only speak the text nodes that have been added or changed.
In a perfect world, a diffing algorithm would be nice, but this would probably be too slow for documents with many updates. Rather we should at least not speak the beginning and end that matches.
This feature is to handle the case where some kind of log keeps appending text nodes.
In the case where it would be instead better to speak an entire chunk of text nodes, the author should wrap them in an element marked as aria-atomic.
Reported by mdcurran on 2015-01-14 05:10
Currently for MSHTML live regions with relevance of text have all text nodes within the deepest element spoken when it changes. We should only speak the text nodes that have been added or changed.
In a perfect world, a diffing algorithm would be nice, but this would probably be too slow for documents with many updates. Rather we should at least not speak the beginning and end that matches.
This feature is to handle the case where some kind of log keeps appending text nodes.
In the case where it would be instead better to speak an entire chunk of text nodes, the author should wrap them in an element marked as aria-atomic.