I'm using the XSSF parser to read a rather large excel sheet with several thousand array formulas. This takes about 30 seconds with the current code, because cell.getRowIndex() is rather expensive and called inside a loop for each array formula. Simply moving that call out of the loop reduces the time to about 3 seconds. The handling is still not optimal (looping through each formula for each cell still), but better already.
Thanks - I've added r1909934. I will look at adding the rest of the PR changes later.
The rest is not that important anyway. Those changes will already help a lot. Thanks!
let's close this then