orga/before-next-major.md
This list contains things that have to be done but will create breaking changes.
Some people set a really big value like 10000000 for convenience. But this will break on some storages or cause really bad performance. Indexed strings should have a limit of like 500 chars because otherwise it makes no sense to index them anyways. Add this check in dev-mode.
This is just another painful setup step. Find a way to not require devs to add the WeakRef types here: https://rxdb.info/tutorials/typescript.html#known-problems
Same for /docs, use a different branch for that.
context string to all write errors [DONE]We already pass the context string to storageInstance.bulkWrite() and when we include it into bulk-write errors, debugging becomes way easier.
https://github.com/pubkey/rxdb-premium-dev/pull/480
https://github.com/pubkey/rxdb-premium-dev/pull/477
toggleOnDocumentVisible should default to true [DONE]https://github.com/pubkey/rxdb/issues/6810
https://discord.com/channels/969553741705539624/1237000453791678487/threads/1327921349808885831
In the RxJsonSchema, a property of a document can have multiple types like
{
type?: JsonSchemaTypes | JsonSchemaTypes[];
}
This is bad and should not be used. Instead each field must have exactly one type.
Having mixed types causes many confusion, for example when the type is ['string', 'number'],
you could run a query selector like $gt: 10 where it now is not clear if the string foobar is matching or not.
Having an .info() method helps in debugging stuff and sending reports on problems etc.
Most "normal" users do not need premium access so we should name it "RxDB Enterprise" to make it more clear that it is intended to bought by companies.
toVersion version field. We should also add a fromVersion field so people could implement performance shortcuts by directly jumping several versions. The current migration strategies use the array index as toVersion which is confusing.