Reported by mdcurran on 2014-01-08 03:59
Currently, NVDA assumes that all radio buttons are direct children of the same parent, and that the only children of the parent are in deed radio buttons. Therefore, NVDA uses the child index and the parent's child count as the 'ordinal' or position information for each radio button.
Clearly this is not at all useful if the radio buttons are within separate panels etc.
We could either 1. no longer expose position info for adio buttons (especially if its 1 of 1), or 2. Come up with a complex huristic that searches for and counts radio buttons around the current radio button, and calculates suitable position information. However, it is debatable as to how high up the parent structure the search should go. Surely there are examples where there are two groups of radio buttons within the same frame?
Therefore, for best performance and suitability for as many situations as possible, we should simply stop exposing this position information for Java radio buttons.
Reported by mdcurran on 2014-01-08 03:59
Currently, NVDA assumes that all radio buttons are direct children of the same parent, and that the only children of the parent are in deed radio buttons. Therefore, NVDA uses the child index and the parent's child count as the 'ordinal' or position information for each radio button.
Clearly this is not at all useful if the radio buttons are within separate panels etc.
We could either 1. no longer expose position info for adio buttons (especially if its 1 of 1), or 2. Come up with a complex huristic that searches for and counts radio buttons around the current radio button, and calculates suitable position information. However, it is debatable as to how high up the parent structure the search should go. Surely there are examples where there are two groups of radio buttons within the same frame?
Therefore, for best performance and suitability for as many situations as possible, we should simply stop exposing this position information for Java radio buttons.