Question about adapting callback to suspend api. I...

# coroutines

[spand]

Question about adapting callback to suspend api. I have the following code below adapting javascript FileReader to suspend. Does this look good?
• Should I cancel the continuation when the underlying resource is aborted (ie. by the browser) ?
• Is this sequence safe ? continuation.invokeOnCancellation -> abort() -> onabort -> continuation.cancel()

suspend fun Blob.readAsDataUrl2(): String {
    val fileReader = FileReader()
    return suspendCancellableCoroutine { continuation ->
        fileReader.onload = {
            val dataUrl = fileReader.result as String
            continuation.resume(dataUrl)
        }
        fileReader.onerror = {
            continuation.resumeWithException(DomExceptionThrowable(fileReader.error as DOMException))
        }
        fileReader.onabort = {
            continuation.cancel()
        }
        continuation.invokeOnCancellation { fileReader.abort() }

        fileReader.readAsDataURL(this) // <-- Async work starts here
    }
}

[elizarov]

Calling cancel from invokeOnCancellation whould not do anything. It is already cancelled.

[elizarov]

So yes, it is safe.

[elizarov]

If you don’t have an extrenal source of abort, then you don’t even have to install fileReader.onabort. But if you do, then you might considering doing continuation.resumeWithException(IWasAborted(…)) there to distinguish cases when a coroutine was cancelled and when something else aborted your file reader.

[spand]

The documentation isnt really clear on if it can happen or not. I guess you are right though that its more of an error.