fastexcel VS docx4j

I'm using fastexcel to create the excel file, and I'm flushing it to the OutputStream after every row is written. Right now, I am using a FileOutputStream to write to disk. When the excel file is done being generated, I read it back in using InputStreamResource and stream the response. My thought process is that a ByteArrayOutputStream keeps everything in memory even if I'm flushing the excel file after every row, so I used the FileOutputStream. Does my logic track here? Or am I unnecessarily slowing things down with expensive filesystem IO?

fastexcel the fastest xlsx generation library

I would use some library like for example this one. So you collect all the data you want to store in a list or map, and iterate over it, then add them to the excel like shown in the documentation of the library (should always be the same way more or less for any library)

I've only had memory problems with apache poi. I only use https://github.com/dhatim/fastexcel now