3 ms·
I'd caution that it's a very "80/20" thing, since the cost/complexity of detecting all "unusual" cases can easily become not-worth-it. For example, if the slo
by Terr_ 2y ago
I'd caution that it's a very "80/20" thing, since the cost/complexity of detecting all "unusual" cases can easily become not-worth-it.
For example, if the slowness of the method depends on non-trivial aspects of the input, or if the performance problem in production comes from someone calling your method too many times on (individually) reasonable chunks of data.