Eli Zaretskii wrote: >> From: Andy Moreton >> Date: Sun, 15 Oct 2017 18:42:27 +0100 >> >> Eli can decide if reverting the previous "unsigned char" -> "char" >> changes is a better fix. > > Looks like going back to unsigned should be cleaner, but I'd like to > hear what Paul thinks. Andy's fixes look good to me. Going back to unsigned would result in several pointer casts that are more dangerous than converting to unsigned. They can be further improved by encapsulating this stuff into a function, and I installed the attached (the second one is Andy's other fix, which I installed in his name). I looked for related bugs in image.c (i.e., caused by my earlier patch) and did not find any.