@powersync/attachments saveToQueue doesn't save additionalColumns

[hunterm-wbc]

The attachments package lets us define additionalColumns when we define our attachment schema.

When we save an attachment in the AbstractAttachmentQueue using the saveToQueue method, passing along the any additional columns as part of the attachmentRecord, it does not save these additional columns, I think this is a bit unexpected.

A work around I think will work is to take the record returned by saveToQueue and perform an update, however this is a bit inefficient, and I think it would be clearer to save any additional columns provided to saveToQueue

[stevensJourney]

My understanding of the approach taken here is that the saveToQueue method (and potentially other persistence methods) would need to be overridden if additional columns are used. I agree that the DX for this is not amazing. In the subsequent implementations of attachment helpers in other SDKs we've removed the ability to specify arbitrary additional columns - replacing it with a single JSON column for metadata.

[hunterm-wbc]

Hi Steven - the metadata approach makes sense. I ended up moving the fields to my record associated to the attachment, and it works decent. If the javascript SDK ever gets the metadata, I'd probably switch to that approach. Thanks for the response!