RFR: 8304831: TextFlow.hitTest.insertionIndex incorrect with surrogate pairs [v8]

Mon May 22 21:45:58 UTC 2023

On Mon, 22 May 2023 17:14:30 GMT, Andy Goryachev <angorya at openjdk.org> wrote:

> what do you people think of the number of pulses parameter? is 10 enough? is there a reasonable upper limit that's needed for any processing to settle?

I think 10 should be a reasonable number of pulses for our tests. Most graphs settle down within 3 layout passes.

> another question is whether it could be made static - i.e. not require a Scene. Perhaps we could add a pulse listener to all visible Scenes? Would that be sufficient, since we only need a signal from one to indicate the pulse has been performed, and there is another and another (... up to 10 or N) pulses that basically guarantees that things are settled.

I doubt this is worth it. This is a test utility, and if one of the very few tests that have multiple scenes needed to call it, it wouldn't be hard to have the test call it for each scene (and would give the test more control).

> And if the layout has not settled by that time, we have a bigger issue (can we detect this? I've seen continuous layout loops in one of my programs)

Right. This is tangential to this PR, but at some point we might want to resurrect the discussion on PR #433 to see if we can improve the current situation.

-------------

PR Comment: https://git.openjdk.org/jfx/pull/1091#issuecomment-1558059567