JIT snapshot has no documentation on training data about sdk

[mraleph]

https://dart.dev/tools/dart-compile#jit-snapshot explains how JIT snapshots are created. "Training data" here refers to the data you give your program during the training run. What that data is depends on the program of course, e.g. if you made a data processing application you should train it by giving large enough input file (or a number of files). The whole idea behind JIT snapshots is that you run your program and then you save machine code that JIT generated while your program was running. This means that the training run should be long enough and "stressful" enough that JIT actually triggers and optimizes hot part of your program.

That being said I would recommend to use AOT snapshots instead of JIT snapshots, unless there is a strong reason to want to use JIT. Meaning that you have benchmarked your application and discovered that AOT is doing really poorly (but in this case it is better to report a bug and we try to figure out a way to fix it).