[19:50:40] I read the latest Newsletter about Abstract Wikipedia. https://meta.wikimedia.org/wiki/Abstract_Wikipedia/Updates/2023-01-19 In this Newsletter I read that the most efficient available implementation based on the Metadata from Test results will be chosen and run in the Backend soon. Do you know how much time it takes to interpret the function and do you think it is possible to in [19:50:40] crease efficiency through saving executables of the implementations in the functions in the backend and then running the relevant one when the implementation is chosen. [20:00:48] The metadata should already be available when you run a function. So yes, we know that. Currently we don't precompile the functions, but I think that's a great idea and certainly something that the system architecture would allow for. I'd support that if someone wants to work on that. [20:08:09] In which category does the interpretation of the function belong to. I looked at the metrics and I saw three categories. Orchestration, Evaluation and Execution. [20:13:53] Depends: compositions and built ins are in orchestration, native code in evaluation and execution. Evaluation is the packaging, execution the actual run.