The actual version of the preview was never loaded and compared to the local version.
Instead, for the preview channel it was incorrectly decided that a new version is available on every update check
* Convert "WebInterfaceManager" to singleton
* Move server webUI mapping to the webUI
* Extract logic into functions
* Retry failed download
* Validate downloaded webUI
* Automatically check for webUI updates
* Add logic to support different webUIs
* Update logs
* Close ZipFile after extracting it
* Add option to disable cleanup of backups
* Ensure the minimum TTL of backups to 1 day
* Schedule the automated backup on a specific time of the day
* Introduce scheduler that takes system hibernation time into account
In case the system was hibernating/suspended scheduled task that should have been executed during that time would not get triggered and thus, miss an execution.
To prevent this, this new scheduler periodically checks if the system was suspended and in case it was, triggers any task that missed its last execution
* Use new scheduler
Some extension require some assets to work properly.
Currently, the extracted jar file does not contain these assets, thus, these extensions wouldn't work
Currently, the "UserConfig" was created in case it was missing.
But in case settings changed (added/removed), an already existing "UserConfig" never reflected these changes and thus, was out of date
* Exclude "default" category from reordering
Due to the "default" category having been added to the database, the index based approach to reorder the categories didn't work anymore.
In case one tried to move a category to or from pos 1, the default category was selected due to being at index 0
* Normalize categories after reordering
Makes sure that the ordering is correct.
E.g. "default" category should always be at position 0
There is a possibility that a partially downloaded file remains in case of an error.
In that case, the next time the image gets requested the existing file would be handled as a successfully cached image.
* Improve default category handling and add cascade to references where possible
* Minor fix for default category
* Make the default category always first in the normalization
