This removes the requirement to modify the aJson library. I have a pull request open against aJson to make bufferless run faster, but it's unlikely to get merged since the person managing the project seems to have abandoned it. Bufferless responses currently causes the code to run unreasonably slow due to the missing optimizations in aJson, but even with them is still slower than the current situation or dynamic buffer. This is due to the large amount of overhead that exists in the HTTP client library when writing data to it piecemeal and the optimizations partially address that by reducing the number of calls into the HTTP client library when possible.
The modifications to SSDP are features that will enable more fun things I have in the pipe, but are completely backward compatible with existing usage.
This removes the requirement to modify the aJson library. I have a pull request open against aJson to make bufferless run faster, but it's unlikely to get merged since the person managing the project seems to have abandoned it. Bufferless responses currently causes the code to run unreasonably slow due to the missing optimizations in aJson, but even with them is still slower than the current situation or dynamic buffer. This is due to the large amount of overhead that exists in the HTTP client library when writing data to it piecemeal and the optimizations partially address that by reducing the number of calls into the HTTP client library when possible.
The modifications to SSDP are features that will enable more fun things I have in the pipe, but are completely backward compatible with existing usage.