the check that was there, screen id not being more
than screen count is not valid anymore with the
screenpool approach, a screen id can be bigger than
the screen count in cases such as middle screen of a 3 screen
system disabled, or driver change (in which unfortunately,
connector names can change)
without this, a new containment was created for each startup.
BUG:369665
created a puroposefully corrupted plasmashellrc file,
where one known screen has a big number as id.
without patch a new containment is added in appletsrc each time
with the patch it behaves fine.
| Automatic diff as part of commit; lint not applicable. |
| Automatic diff as part of commit; unit tests not applicable. |
I can see how this is a bug that needs fixing for the a screen ID > screenCount but what about the case where the screenID isn't valid anymore?
i.e
a panel was on screen 2 you shutdown (assuming ID's are also 1 and 2 respectively)
when you boot up you now only have 1 screen
Before the code would return -1
This would put it on screen 2 which doesn't exist - will that cause a problem?
i think it shouldn't (as, no views should be created anyways unless we do actually have a qscreen for it)
but it definitely needs more test.
i think this should either require more checking when creating panels (only panels i think, desktops may have problems)
or, change the check to see if an association between screen id/name is in screenpool and if there is a qscreen corrisponding it.
I probably like more the latter as it would have almost the same logic as before, just "correct"
this version has the old logic again, it iterates trough all the screens but doesn't check for ids being < of the count, it just checks that thre is a qscreen for that screen id by using the association in screenpool