I appreciate your wish to keep all beta tests in step. (In many years of beta testing on several applications, On at least one occasion a tester reported a fault which turned out to be in a previous beta and corrected in the current. Some of us wasted time trying to reproduce it!)
However, I did have a problem with the upgrade to beta8. I did not use Scrivener for most of August and into September so when I did use it had to download and install the new beta. Unfortunately, I had not exported the settings and could not do so as the previous version had expired. I took the risk of installing over the previous version So far I have not seen anything going wrong as a result, but if there is a catastrophe I will:
1 Export the settings. (If not possible, I can restore an image backup of the system volume).
2 Uninstall Scrivener and delete all its files. (I am not sure whether I should prune registry entries).
3 Create a new installation of the current beta.
4 Import the settings.
5 If (as is likely) the project files are corrupt, I will restore the last good one available.
Might it be possible to give the installation process an option to recover the settings from a previously installed version and incorporate them in the new one? This might not just be a beta feature, since it would be a useful safeguard for someone carrying out a normal upgrade who forgot to export settings.
We use a Windows server but not its native backup (instead we use urbackup, an open source third party application). No working data is held on any computer’s system disk and these disks have an image backup daily Data is stored o Server folders, including redirected documents, This enables all users to access their own data when logged in to any machine. Especially for laptops,the most used folders (e.g. documents) are made available off-line so they can be accessed even if not connected to the home network.
With this backup strategy I can not only expect to cope with the catastrophe described above but also feel confident that I will not suffer any loss if the beta malperforms excessively.