OpenXml (xlsx) out of memory with millions of hidden rows...

Created by: Jim Hargrave

This file has a sheet with over 1 million spreadsheet rows with only 93 of them as visible. So when okapi is doing the "merge" step, it's trying to hold all of those in memory which means holding the java object representation of data for those hidden rows and then rendering those as XML and storing all of that in memory before writing out the rendered XML to the file.

I assume the hidden rows are all being stored in a large object and kept in memory. Is there a way to break these up so they can be garbage collected more efficiently?