Hey,
as part of the rewrite of the Persistent Storage settings (#17803), I
also want to solve #11529 "Save data to Persistence when it is created
(no need to restart)".
The only way I see to solve this while providing a nice UX is to rewrite
part of live-persist (and the live-boot functions uses by it) to allow
activating/deactivating a single persistence feature (that's the term we
want to use instead of "persistence preset" IIUC).
This is the user flow I envision:
* In the Persistent Storage settings GUI, the user clicks on the switch
of a persistence feature to activate/deactivate it.
* The frontend calls the Activate()/Deactivate() method of the feature's
D-Bus object.
* The backend checks if any processes are running that must not be
running when changing this feature (for example for the Thunderbird
persistence feature, no process with executable "/usr/bin/thunderbird"
or "/usr/lib/thunderbird/thunderbird" must be running).
* If any such process is running, the backend sends a signal that it's
waiting for these processes to exit.
* The frontend receives the signal and displays a message to the user
that they have to close the corresponding app ("Thunderbird") to
continue.
* Once all conflicting processes have exited, the backend
automatically continues activating/deactivating the feature.
* The backend mounts/unmounts the files/directories of that feature.
* The backend adds/removes the corresponding line(s) to/from
persistence.conf.
live-persist is not able to mount/unmount a a single file/directory
instead of the whole persistence.conf. That's why we need to rewrite
part of it.
I plan to do that in bash, copying the parts from live-persist and
live-boot which we need for that.
During boot, we could still use live-persist, or we completely replace
it with the new script (which means that it should also be able to
activate features from a config file).
What do you think about that plan?
Cheers
