A simple caching system to improve your workflow

Normally, every time you run your script, everything gets calculated again, even the parts of the script that did not change This package solves this waste of calculation time, by using reflection to generate a dependency graph of your source code:

Simply add @cached in front of a function name like in example.jl:

function depend(x)
    return x+1
end

@cached function heavy_calculation(x)
    println("doing heavy work work for x = ", x)
    sleep(2)
    return x * depend(x)
end

println(heavy_calculation(1))
println(heavy_calculation(2))
println(heavy_calculation(1))

Now, heavy_calculation will only run twice, (once for x=1 and once for x=2) instead of three times. If you restart Julia, heavy_calculation won't run at all. Instead, heavy_calculation(1) will return a cached result.

If you change the source code of heavy_calculation, or depend, then heavy_calculation will run again if you execute the program.

If you change the source code of a function that is not used by heavy_calculation, then heavy_calculation will not run again if you execute the program.

Note that the way that SimpleCache.jl uses reflection to see which source code is relevant to a function is written in an extremely hacky way. Therefore, there currently are dependendies that SimpleCache.jl won't be able to detect: Uses of eval and conditional function pointer usuage.

Also note that SimpleCache.jl will never delete any cached data from disk. You will have to manually delete it if your disk storage runs full.