Paul Eggert wrote: > - bool unibyte_locale = MB_CUR_MAX == 1; > + bool unibyte_locale = USE_C_LOCALE || MB_CUR_MAX == 1;
I think this part causes a regression on platforms where the "C" locale uses UTF-8 encoding (macOS, Haiku): Before the patch, an argument that contains non-ASCII UTF-8 characters was displayed in full. After the patch, an argument that contains non-ASCII UTF-8 characters is displayed with escape characters for the non-ASCII bytes. Bruno
