moving to a DB meant empty slices in the Configuration & Policy structs
were being stored as null, and striking a nerve with Jellyfin.
Mediabrowser library change fixed that by de-nulling them itself, and a
new bool field called "Homescreen" is now used to decide if a profile
has a homescreen layout stored or not. This field is hopefully correctly
filled in during migration.
ombi_template, configuration, displayprefs, and policy still stuck
around for the admin new user feature. They are now sourced from the
default profile, and eventually a feature to select the source (or no
source) will be added.
this was still used when creating a new user as admin for some reason.
template is now sourced from the default profile.
don't think there's a way to negate a query with badgerhold, so i can't
do "delete(not (where JellyfinID in <ExistingUsers>))", and the old
method of rebuilding the store is no longer possible.
migrating to badger, with the badgerhold frontend. So far, done:
* Announcements (small, for a quick test)
* Discord/Telegram/Matrix/Email
most interaction with badgerhold is done through the standard
Get<x>/Get<x>Key/Set<x>Key/Delete<x>Key. UserExists functions have been
added for email and matrix, and those and the original ones now use a
query against the database rather than sifting through every record.
I've tagged these searched fields as "index" for badgerhold, although this
definitely isn't used yet, and i'm not entirely sure if it'll be useful.
migrateToBadger is now in migrations.go, and a temporary config key
"migrated_to_badger" has been added, although it isn't being used yet,
migration is just running every time during development.
Low German/Saxon (NDS) is empty entirely, which caused discord lang
registration to error, so i've just filled in the name. Somebody
directly translated "English (US)" into italian instead of putting
Italian in italian, corrected that. Use some common sense!
functionality not done yet, just comitting here because there were lots
of adjustments to layout stuff, accomodating for most combinations of
card presence/size.
the user shouldn't know if the reset has actually been sent (i.e. if an
account with the given contact address exists), so the backend response
is always sent after 1 second.