Inaccessible/inescapable UI when restricted to keyboard navigation about imgui HOT 3 CLOSED

You're right, that works, my bad. Should've looked up basic controls. As you might have guessed, I don't normally use just the keyboard, I was testing the feature. Just talked with a UI Engineer about Dear ImGui and its use for desktop applications like mine, and was curious how well supported more niche use cases are (was playing with the idea of eventually integrating screen readers, this makes it a lot more realistic).

I'll definitely have to tweak my application a bit to make it work well. E.g. I have a log section with each entry being a selectable and a follow state for new logs - switching between that follow state and navigating logs is turning out to be quite annoying. More research on my part needed as to what how other applications handle such an interaction.

Thanks for your help, and sorry for not researching this properly.

from imgui.

Have you tried tapping Alt to switch to the menu layer? Both of your problems should be solved by this :)

It’s the standard input in Windows land to do it. If you didn’t discover it, i am curious as to what input you would normally use on your system to do that?

from imgui.

I double-checked and it is mentioned in the ShowUserGuide():

Also linking to #4122 about screen readers. I don't know much about them, and I think it may be difficult to meet all standards, but I am certain we could implement some base line features such as describing newly focused item while tabbing etc. We should need to be vary we are likely to be hitting some walls as dear imgui tends to not have enough holistic understanding of the UI setup, and some items may be lacking semantic, or dissociating focused item/identifier from e.g. labels. Typically when you use a row-spanning Selectable() which is the item being interacted, it's a Selectable with no label, with some description separately.

I plan at some point to rework the "capturing" LogXXX api into a nicer API and we can probably push improvements there.
We also had an experimental feature (not currently pushed) in imgui_test_engine that extracted text and colors directly from vertex data. by reverse mapping UV coordinates to characters, which has its use for testing but has the limitation that any clipped text won't be seen by it (in a way, it would "see" the same thing as a user, but reading a clipper text by clipping the actual prononciation is not going to feel great..).

from imgui.