I performed a number of Google searches on this, but could not find anything (other than confirming that AppEngine ignores calls to "resp.setHeader" ("transfer-encoding", xxxx). "
My problem is that I have a server process that takes a little time to complete (~ 3second avg), but which gives results in a continuous stream (i.e. the first data object is ready in only 10 ms).
Currently, the server buffers the response and then spills out the entire lot to the client for display. Although this works, it is not a brilliant user interface, as the user has to wait for the completion of the entire process before they can see the first results ...
I believe that with the help of batch transmission this would solve, since I could immediately send the first response to the client (and transmit the subsequent data as soon as it becomes available), giving a much better idea of ββprogress, etc.
My problem is that I cannot decide how to enable packet transmission.
The Google engine for switching channels and the application engine mainly displays messages from people who want to disable chunked transfers from blobstore, or messages that indicate that channel transfers are enabled by default. However, my testing to date has shown that, at least for my application, this is not true.
Likewise, non-application links talk about setting the transfer-encoding content header, but this is explicitly ignored in AppEngine (at least this is according to python docs - nothing was found on the Java side, this clearly indicates, but it doesn't It worked when I tested it).
Finally, I already tried to clear the repos object, get the responder and flush, get the response output and flush, etc. it does not matter.
So - my question is: how to enable batch migration?