> Are we talking about a data scientist performing initial, exploratory analysis on a very small subset of data? Yes, that's one example. Also when debugging o

> Are we talking about a data scientist performing initial, exploratory analysis on a very small subset of data? Yes, that's one example. Also when debugging one usually uses small data sets. There are plenty of cases where runtime is short. I think the problem is that Julia is somewhat vague on how it should be used. If it stated explicitly that it is intended to be used in MATLAB-like fashion with one long-running instance that would save people from trying to use it as Python or other dynamic language.