https://bugs.documentfoundation.org/show_bug.cgi?id=168409

--- Comment #17 from Michael Weghorn <[email protected]> ---
(In reply to Jean-Pierre Ledure from comment #15)
> > Does that sound like a better potential approach/workaround for now?
> 
> It is not better without the "Rect" property.
> Otherwise, if there is a reasonable expectation to have a +/- stable UITests
> API for a foreseeable future, then we can give it a try.

Regarding UITest API stability, Xisco might possibly have more insights on
that.

(In case of using UITest API, adding a UI test that uses this might be nice:
Even if we don't necessarily guarantee stability, this will at least make clear
when something changes + can remind of notifying you/the ScriptForge developers
if that's the case.)

> Tell me what your preference is.

If that's a viable option for now, I'd prefer this instead of using a11y API,
as I'm expecting to make more changes to a11y in the future.

> What would be the delay ? We should provide a correction in a forthcoming
> 25.2 release. Note that I can make a patch as described in comment#13 within
> a week or so.

https://gerrit.libreoffice.org/c/core/+/191456 is a demo for a potential
implementation. If you're building LO from source, you could try whether you
get all information as needed this way.

I went with 4 properties "X", "Y", "Width" and "Height" instead of "Rect",
because the UITest code always uses OUString for the value, so I think having
one property for each value might be nicer than having to encode the whole rect
as one string + then parsing the single values from that again on the
ScriptForge side. But what the best format would be is totally open for
discussion, if that's the way we want to pursue.

One option could also be to implement the solution from comment 13 right away
for master and 25-8, then switch to the UITest API based solution for master
only once it's ready, without the need for it to be finished before the next
25.8 release.

-- 
You are receiving this mail because:
You are the assignee for the bug.

Reply via email to