You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
All the kids in the block are doin' it, but we want to do it right™.
Resumable uploads require a FileReader object so only work with JS. As such, we can implement the REST backend only, with JS support augmenting the current "autoprogress" form wrapper.
Basic concept:
Login or grant ID is required to create an upload session id.
To open the session, you need to announce file list, file sizes and file names.
Once initialized, an upload session pre-allocates the storage
The REST api should allow to write a chunk of binary data given the triplet: upload ID, file index, byte range.
The upload session should have a very short TTL (~30m idle max)
We can then modify the web forms and REST interface so that file uploads can either accept form uploads or an upload session ID. If the upload SID is given, we just move the files into place and handle it exactly as it's being done now.
On the client side, if FileReader is supported, we remove the object from the DOM and handle the upload outself as above, by chunking the files into small blocks and uploading them using vanilla POST requests. The server is oblivious about the chunking, as long as the addressing triplet is valid. Once the files are uploaded, we inject an hidden input field in the form with the upload SID and proceed normally.
The text was updated successfully, but these errors were encountered:
All the kids in the block are doin' it, but we want to do it right™.
Resumable uploads require a
FileReader
object so only work with JS. As such, we can implement the REST backend only, with JS support augmenting the current "autoprogress" form wrapper.Basic concept:
We can then modify the web forms and REST interface so that file uploads can either accept form uploads or an upload session ID. If the upload SID is given, we just move the files into place and handle it exactly as it's being done now.
On the client side, if FileReader is supported, we remove the object from the DOM and handle the upload outself as above, by chunking the files into small blocks and uploading them using vanilla POST requests. The server is oblivious about the chunking, as long as the addressing triplet is valid. Once the files are uploaded, we inject an hidden input field in the form with the upload SID and proceed normally.
The text was updated successfully, but these errors were encountered: