https://gcc.gnu.org/bugzilla/show_bug.cgi?id=128107
--- Comment #3 from Juno Krahn <juno.krahn at gmail dot com> --- Not using AI to track bugs. I was getting odd results in a particular code and by the time I tracked down the precise condition my notes were rather long, and I admit to using AI to clean up my notes which I should have done myself. Brief summary: On a file with STREAM, FORMATTED, ADVANCE=YES, and reading an integer array for a short data input (i.e. last line of a data section) then for every integer attempted to read past available input data, file position gets incremented by one, and those persists through remaining reads. Earlier today i found another clue: I suspected EOL record processing, so I tried a file with CRLF input and this seemed to make the POS incrementing bug disappear. I thought in my original project code that PAD= setting seemed to make a difference in some cases but I now can't reproduce it.
