SUGGESTION: Allow specifying a box file_id in box_write

[alexbrodersen]

Currently box_write requires a file_name and dir_id.

The API documentation suggests it's possible to specify a file_id instead.

It seems non-symmetric that box_read can handle a file_id but box_write cannot.

Let me know if you would like me to attempt a pull request to solve this.

[nathancday]

box_write() is a convenience wrapper around box_ul(), which uses this API endpoint. Today box_ul() handles both uploading new files and updating existing files (which are slightly different requests). You are correct that you can update with a file_id. But you can't upload a new file by a file_id.

Would you want to rework box_ul() so it would decided whether to update or upload new based on the presence/absence of file_id = or create a new function like box_update(file, file_id)?

I'm skeptical about changing the behavior of box_ul() because it's core function not only for users but other boxr functions like box_push(). What do you think?

[alexbrodersen]

I think a new function box_update(file, file_id) would be best.

Then, would it be possible to use box_write to dispatch to the correct method based on if a file_id or a name as been provided?

[nathancday]

I think that sounds nice, something like if (!is.null(file_id)) box_update(...). I'd welcome a PR for review.

I'm not sure name will work this , because at the API level names aren't unique enough. I know box_push generates a directory diff to emulate the by-name look up, but that starts to feel like a different problem. What do you think?

FWIW I've always saved objects locally and relied on box_push to sync local/remote, could that be a solve/hack in your workflow?

[ijlyttle]

This sounds like it could be a part of #208.