Analysis continues on old content if file content is replaced by unrelated code

[MikhailArkhipov]

Start with

import plotly

when analysis starts running, paste over code from #960

import sys
from collections import defaultdict

def xx():
    dmap = defaultdict(list)
    sizes = {db:int(sz)//(1 << 20) for sz, db in [ln.split() for ln in open(sys.argv[1]).readlines()]}
    return {srv:sum([sizes[d] for d in dbs]) for srv,dbs in dmap.items()}

Observed: analysis never completes. If there is progress it stops at the same number of items. Or it continues analysis plotly

[AlexanderSher]

Yeah, we aren't unloading modules that aren't referenced anymore. Should we do this btw?

[MikhailArkhipov]

Technically yes since otherwise memory consumption will be growing indefinitely. Not super urgent though.