We warn people in the pods that it is not a good idea to mix locale and utf8. I propose adding a warning message when it does happen. What actually goes on in this circumstance is that the actual locale is ignored. If that locale is not ISO 8859-1, then your results are likely to be garbage to whatever extent the locale differs from 8859-1. The 'A' in your locale may, for example, not be Unicode's 'A'. Since 8859-1 is a subset of Unicode, that locale will result in non-garbage, but you may get unexpected results with characters matching outside the locale.