Okay, full disclosure, this happened at least 5 years ago on a computer and printer that are looooong gone. You will not get any information beyond what I offer in the OP because I simply don’t remember.
As far as I recall, this happened exactly twice. Both were PDFs, but other than that had no linking factors (other than having text instead of embedded pictures of course). It did not occur on the actual computer display, and only happened when printed – meaning it probably occurred during spooling or something.
What happened was, simply, every character was sequentially shifted to the next character, meaning A->B, a->b, c->d. There were no zs in the original texts, so I’m not sure what would have happened there. Does anyone have any idea what might cause an entire document to shift every character exactly one character forward when being fed to the printer?
This was on Windows XP probably, the printer was probably one of those scanner/copier/fax Canon printers, not sure on what model.
Nah, one of them was a band practice sheet that I printed out monthly and it always printed correctly otherwise. In fact, when I saw what happened I printed it again and it printed fine (though I did turn in the copy with the character shift as a joke).
Obviously those characters are consecutive in pretty much any character set, so it’s conceivable that some stray increment operation could be happening - perhaps only occurring under very odd and specific circumstances such as an unhandled stack overflow where a program (i.e. the printer driver) accidentally overwrites part of its own code in memory.
I’d expect an error of this sort to be accompanied by a random assortment of other kinds of occasional problems though - failure to print, unreadably corrupted output, etc.
Even those that are not derived from ASCII still generally assign the alphabet adjacent values in alphabetical order. There’s not really a good reason to do otherwise, and doing so has several advantages.
Sure - that’s why I said ‘pretty much any’, not ‘each and every’.
In any case, it seems highly likely that such a character set was in use in this scenario (at least at the point of failure), just because of the described symptom - consider how unlikely it would be for alphabetic output to be shifted by one and only one letter in a character set where the letters aren’t consecutive - it’s possible, but it would indicate a designed error, rather than accident.
I imagine those that don’t are the ones that were designed from the viewpoint of something other than the Roman 26-letter alphabet, and without any intent of being used for Western/Roman text.