which was done in 2015 to stop kwallet dunning in the live session completely.
This sadly now has the consequence that plasma-nm sees no actual written config disabling kwalllet, but maybe simply fails to access it thinking it should be available, then falls back to the password with network manager. Hence possibly the double password request.
Since 2016 releases of plasma 5.7.x the change "Store password into NetworkManager by default when KWallet is not enabled" enables fallback to network manager, but seemingly only when plasma-nm knows kwallet is disabled by config, not when it has been just broken by casper.
So one possible solution could be to leave kwallet's files in place, but write a temp config file 'kwalletrc' for just the live session disabling kwallet, so plasma-nm knows to not use it, hence using a working route first time.
Seems there is still another aspect to this, resulting in it still being unfixed for the Kubuntu Live session.
casper runs in scripts/ casper- bottom/ 34disable_ kde_services
rm -f /root/usr/ share/kde4/ services/ kwalletd. desktop share/kservices 5/kwalletd5. desktop share/dbus- 1/services/ org.kde. kwalletd5. service share/dbus- 1/services/ org.kde. kwalletd. service
rm -f /root/usr/
rm -f /root/usr/
rm -f /root/usr/
which was done in 2015 to stop kwallet dunning in the live session completely.
This sadly now has the consequence that plasma-nm sees no actual written config disabling kwalllet, but maybe simply fails to access it thinking it should be available, then falls back to the password with network manager. Hence possibly the double password request.
Since 2016 releases of plasma 5.7.x the change "Store password into NetworkManager by default when KWallet is not enabled" enables fallback to network manager, but seemingly only when plasma-nm knows kwallet is disabled by config, not when it has been just broken by casper.
So one possible solution could be to leave kwallet's files in place, but write a temp config file 'kwalletrc' for just the live session disabling kwallet, so plasma-nm knows to not use it, hence using a working route first time.