After #2231 is done, compare the XPRA_LEGACY_FRAME_PER_CHUNK=1
mode with the default behaviour when connecting with both the python client and (more interestingly) with the html5 client.
I would expect the difference to be minimal: only measurable as fewer bytes / raw packets, but who knows.. At least it will help confirm that we have a correct test baseline.
Need test hardware to run the tests.
this ticket has been moved to: https://github.com/Xpra-org/xpra/issues/2232