One other problem/feature I noticed on AROS. AROS has different semantics w.r.t MUIA_Application_Sleep and newly created windows.
There are 2 differences which I in the following I will call
open-window-only and
sleep-underflowopen-window-only: In classic MUI (m68k, AOS4, MorphOS) one can create a new window, set the application to sleep and then open the new window, which will not be put to sleep (but all other windows are). On AROS though the newly created window is also put to sleep. Classic MUI only sets open windows to sleep whereas AROS seems to also set hidden/closed windows to sleep.
It is easy to work-around this by first setting the application to sleep before creating and opening the new window.
sleep-underflow Classic MUI allows to unsleep the application and windows that weren't put to sleep are still in their no-sleep state. Said differently, the sleep counter of a window does not drop below 0 when calling MUIA_Application_Sleep, FALSE. On AROS though it seems this sleep counter underflows, thus when creating and opening a new window and unsleeping the application, the new window is set to sleep. It gets even more weird when calling MUIA_Application_Sleep, TRUE as this can make a window to get a zero sleep counter and allow user input.
It is possible to work around this by making sure that between every call to MUIA_Application_Sleep, TRUE and MUIA_Application_Sleep, FALSE, no new window is created. But it is not a trivial one. Often it is just very convenient to set the whole application to sleep, do whatever changes are necessary, which might include adding new windows, and then unsleep the application.
I just did an audit of all sleep/unsleep calls in MUIbase in order to match the semantics of AROS (see e.e.g this commit:
https://sourceforge.net/p/muibase/code/3918).
MUIbase is probably not the only application that runs into this issue, thus, my recommendation would be to change the semantics to how classic MUI does it. And BTW, both GTK and Qt also work the same way as classic MUI. In particular the sleep-underflow change would be highly desirable.