CODEWITHSUNDEEP
pythonpython-3.xcachingoptimization
Aart Goossens
Aart Goossens

Reputation: 283

Python lru_cache usage optimization

Say I have this method that I cache using lru_cache:

@lru_cache(maxsize=8)
def very_expensive_call(number):
    # do something that's very expensive
    return number

I am calling this method like this:

print([very_expensive_call(i) for i in range(10)]) # call_1

Because the maxsize of the cache is 8, only numbers 2-9 are cached at this point.

After call_1, I am doing call_2:

print([very_expensive_call(i) for i in range(10)]) # call_2

During call_2, again first number 0 is called (not in cache!), and after that numbers 0 and 3-9 are cached. Then number 1 is called (not in cache!) and after that numbers 0-1 and 4-9 are cached. Well, you see where this is going: the cache is never used...

I understand that for this specific example I could alternate between range(... and reverse(range(... but in a more complicated scenario that's probably not possible.

Question: Is it possible to inspect which numbers are cached and to order the calls based on that? What would be the overhead for this?

Upvotes: 0

Views: 677

Answers (1)

Dimitris Fasarakis Hilliard
Dimitris Fasarakis Hilliard

Reputation: 160417

No, no, the cache used in lru is specifically designed to not be public-facing. All its internals are encapsulated for thread safety and in order to not break code if the implementation changes.

Apart from that, I don't think it is a good idea to base your input based on caching, you should cache based on your input. If your callable is not periodically called with the same arguments, maybe a cache is not the best option.

Upvotes: 2

Related Questions