Is there a Multiplatform library that allows me to...

# multiplatform

[Christian Gaisl]

Is there a Multiplatform library that allows me to persistently store large(-ish) ByteArrays (a few MBs) and have a simple key-value Flow-based API to check wether that key exists or the ByteArray under that key has been modified?

[Michael Paus]

Which (multi)-platforms? Something that works on desktop for example does not necessarily work on WASM too.

[Christian Gaisl]

I'm mostly looking for a mobile (Android & iOS) solution

[Ahmed na]

sqldelight uses sqlite internally which supports blob

ideally should support your use case

[Christian Gaisl]

@Ahmed na I've also tried Room for my use case. Room or SQLDelight with BLOBs sounds like the thing I'm looking for, but in practice I ran into the issue that with BLOBs larger than 1MB i get an error telling me the SQLite CursorWindow size is too small.

After some research, there seems to be a practical limit of 1MB for BLOBs in the underlying SQLite on Android, wether you use Room or SQLDelight.

Also, for my use case, I don't necessarily want to query the whole ByteArray everytime something changes, I just want to be notified of changes. For a given key I want to be notified wether a ByteArray for that key has been created, deleted, or edited. Depending on circumstances I want to then be able to choose wether to load the ByteArray into memory. And all of on Android & iOS at the same time, if possible.

Thinking about it, I think I'm looking for a simple Multiplatform FileObserver thing. Does something like that exists?

[Ahmed na]

@Christian Gaisl file sdk will be kotlinx io
but so far dont know if any filechange observer is there
--------

One hacky idea for the db approach is storing the byteArray on multiple rows with same key name prefix

file_a_1 -> [0-100]   ;1MB
file_a_2 -> [101-200] ;1MB
file_a_3 -> [101-200] ;0.5MB

then query back the file where like = file_a%

and merge into a ByteArray

[Christian Gaisl]

Thanks, I'll have a look at that 👍

[Michael Paus]

This limit of sqlite can be easily overcome. Just set the window size according to your needs. I am storing about 2GB in sqlite and most single blobs are much larger than 1MB.

actual class DatabaseDriverFactoryImpl actual constructor() : DatabaseDriverFactory {

    private lateinit var activity: ComponentActivity

    fun init(activity: ComponentActivity) {
        this.activity = activity
    }

    override fun createDatabaseDriver(dbFilePath: String): SqlDriver {
        return AndroidSqliteDriver(AIPDatabase.Schema, activity.applicationContext, dbFilePath, windowSizeBytes = 50 * 1024 * 1024)
    }

}