After some more investigation, in the non-rootless mode, Xmir sets the _MIR_WM_PERSISTENT_ID atom on the root window, and since pasted just queries the default root window, this works fine.
However, in rootless move, one of the children on the "root" window gets the _MIR_WM_PERSISTENT_ID atom, thus pasted is not able to get the persistent surface id.
So, we can approach this one of two ways.
1. Xmir can set _MIR_WM_PERSISTENT_ID on the root window regardless of the mode.
2. pasted will have to get smarter and try to first query the root window and if that returns nothing, then do an XQueryTree() and iterate over child windows until the _MIR_WM_PERSISTENT_ID is found.
I prefer 1 since it will offer consistency, bit I'm not sure if there are any pitfalls with that approach. Will subscribe xmir to this bug and get duflu's advice:)
After some more investigation, in the non-rootless mode, Xmir sets the _MIR_WM_ PERSISTENT_ ID atom on the root window, and since pasted just queries the default root window, this works fine.
However, in rootless move, one of the children on the "root" window gets the _MIR_WM_ PERSISTENT_ ID atom, thus pasted is not able to get the persistent surface id.
So, we can approach this one of two ways.
1. Xmir can set _MIR_WM_ PERSISTENT_ ID on the root window regardless of the mode. PERSISTENT_ ID is found.
2. pasted will have to get smarter and try to first query the root window and if that returns nothing, then do an XQueryTree() and iterate over child windows until the _MIR_WM_
I prefer 1 since it will offer consistency, bit I'm not sure if there are any pitfalls with that approach. Will subscribe xmir to this bug and get duflu's advice:)