The open() variant reads the first 100 text characters, not 100 bytes, because open() works in “text mode” by default, so the two are not _that_ different. Your

The open() variant reads the first 100 text characters, not 100 bytes, because open() works in “text mode” by default, so the two are not _that_ different. Your generally point is still very correct though, the memory issue is the main problem with the “suggestion” here.