Could you please elaborate a bit on the new "resumeheader" parameter?

Could you please tell me what "HttpClient" version is JFileUpload 2.6 currently using?

After analyzing the network traffic it seems that JFileUpload is treating the upload differently. At a TCP level, the server is sending an "ACK" after every two frames sent by the client. That is after every ~2KB. No wonder it slows down the transfer speed so much. This is most probably related to t...

You can also see how JFileUpload's speed compares to a Flash based uploader. On the left is JFileUpload and on the right is the Flash uploader.

My measurements were not good after setting the parameter. The transfer speed improves but is still far from what you can get using the classic HTML upload. You can see the two methods compared in the attached graph. Could you please tell me why is JFileUpload applet so much slower that any browser,...

Setting the value of "retryonerrorheader" parameter to "false" it will make the applet stop retrying when the server replies with the defined "errorheader" HTTP header. However, in JFileUpload version 2.6, this makes the JFileUpload cancel any retries also if the error received is org.apache.commons...

Is the following warning related to the "buffersize" parameter that I've set?

WARNING: Going to buffer response body of large or unknown size. Using getResponseBodyAsStream instead is recommended.

This feature is working great so far, but it will be really useful for the user if the progress bar will change based on what chunk is currently uploading. If you upload 50% of a file and for some reason you were interrupted, dragging the file again it will show the progress bar from beginning (and ...

I experience this under Windows, but it seems to be worse under Mac OS X.
Setting the buffer size parameter helps a lot. Now it's almost as fast as the standard HTML upload, so thank you for the solution.

This is quite important, as we have a large number of GB files to upload and the transfer speed difference will count a lot.