[WIP] Support for OPFS #841
Conversation
There was a problem hiding this comment.
This type also defines methods that return async iterators, which don't appear to be supported at the moment in scala-js. It makes sense to me to tackle those methods in a separate piece of work. I've asked about potentially adding support over in the scala-js discord.
There was a problem hiding this comment.
| * FileSystemFileHandle.createWritable() method. | ||
| */ | ||
| @js.native | ||
| trait FileSystemWritableFileStream extends js.Object { |
There was a problem hiding this comment.
I'm a little lost here at the moment. This should actually extend a WritableStream type, and I started implementing that before realising that there's already a WriteableStream (sp) defined in scala-js-dom. However what's already defined looks quite different to what I'd have expected from the whatwg/MDN spec. I came across #754 which looks related.
Ultimately I'm not sure how to proceed with the facade for FileSystemWritableFileStream when it comes to whether or not to integrate with the existing WriteableStream. I'd appreciate any thoughts or advice.
There was a problem hiding this comment.
However what's already defined looks quite different to what I'd have expected
In what way? There are a couple mistakes but it should be the same thing.
There was a problem hiding this comment.
My bad - definitely should have elaborated in the original comment.
My confusion in primarily that the existing interface in scala-js-dom seems to resemble the spec for WritableStreamDefaultWriter rather than WritableStream. However the current scala-js-dom version also includes a state method that doesn't exist in either public interface (though state is an internal slot for the WritableStream spec, albeit with different potential values).
If we treat WriteableStream as a misnamed WritableStreamDefaultWriter, then it seems clear that an implementation of FileSystemWritableFileStream should not extend WriteableStream. However we have methods like CompressionStream#writable that would then be returning WritableStreamDefaultWriter rather than an actual WritableStream like the spec suggests.
Maybe the spec has evolved over time, or I'm missing a trick with WritableStream = WritableStreamDefaultWriter.
There was a problem hiding this comment.
Hmm, I see. That's a quite a bit more than a couple mistakes 😅 thanks for bringing that to our attention.
Since the name doesn't correspond to anything in the actual DOM maybe we should just deprecate it and start over, with the right names, class/trait designations, and APIs 🤔
There was a problem hiding this comment.
Thanks, glad to know I'm interpreting the code correctly. I'm still familiarising myself with a lot of this stuff. :)
Open to any ideas you guys have on how to handle this. Maybe it's best if I remove FileSystemWritableFileStream from this PR to simplify things, and we can move discussion of what to do about WriteableStream to a separate issue - maybe #629?
Hi!
I was hoping to add support for the OPFS. I've run into two areas where I was hoping to get advice - I'll highlight these in PR comments.