I am glad to hear that. In fact we are just starting playing with BP, so I cannot (yet) suggest any precise API for it because I am still learning. I guess one of the approaches would be to create (virtual) file handlers similar to a way that you upload altered images. Those could point to a specific file portion (with offset and length), and you could easily use them with the Uploader.
Another thing is Uploader - server communication, because here a semi-protocol should be designed, but I guess you could leave it for developers. It should be really easy once you can access files in chunks. What could happen for a single (large) file is:
- Client initializes the upload, it uses a POST to a server with just some meta-data about a file (size, how many chunks)
- Server provides unique URL for this upload
- Client uploads chunks, one by one.
- Client can ask Server about the status - e.g. which chunks are missing
What you really need from BP here is uploading in chunks. The above case could also allow pause + resume of uploads, handling failed chunk upload etc.
yay. so now that 2.2.0 is out I get to make promises. here's my promise. I'll get FileAccess on github by the end of next week. and start hacking in chunking support out in the open. all help is welcome.
Next. As far as upload we've long envisioned at least 2 services. the first one, "Uploader" was designed to be a client side drop in replacement for existing forms. so that people wanting to integrate wouldn't even have to change their server side logic that recieves the upload.
The *next* version would be something deep and custom, and more efficient. That thing could indeed be an exploration into new ideas and perhaps standards around file upload.
The Uploader service should be on github in a month or so, and at that point we can start implementing something that is truely targeted at resumable and efficient uploads.
Any pre-existing work we can leverage here?
best,
lloyd
