I have a post looking at Polybase support for different compression formats:
This is a very interesting set of results. First, 7Zip archived files do not work with the default encoding. I’m not particularly surprised by this result, as 7Zip support is relatively scarce across the board and it’s a niche file format (though a very efficient format).
The next failure case is tar. Tar is a weird case because it missed the first row in the file but was able to collect the remaining 776 records. Same goes for .tar.gz. I unpackaged the .tar file and the constituent SecondBasemen.csv file did in fact have all 777 records, so it’s something weird about the codec.
Stick to BZip2 and GZip if you’re using flat files.