On Thu, 2012-06-21 at 22:34 +0200, Ole wrote: I don't quite understand the rationale behind the separator characters you have chosen. Could you explain them? The textarea_select_fragment function looks wrong to me. There seems to be some confusion between character indices and byte offsets. The following should work better: static bool textarea_select_fragment(struct text_area *ta) { static const char *sep = "/.\r\n"; int c_caret_pos, index; size_t b_start, b_end; int c_sel_start = 0, c_sel_end = ta->text_len; /* Get character index of caret */ c_caret_pos = textarea_get_caret(ta); if (c_caret_pos < 0) { return false; } /* Compute byte offset of caret position */ for (b_start = 0, index = 0; c_caret_pos-- > 0; b_start = utf8_next(ta->text, ta->text_len, b_start), index++) { /* Cache the character offset of the last separator */ if (strchr(sep, ta->text[b_start]) != NULL) { c_sel_start = index; } } /* Search for next separator, if any */ for (b_end = b_start; b_end < ta->text_len; b_end = utf8_next(ta->text, ta->text_len, b_end), index++) { if (strchr(sep, ta->text[b_end]) != NULL) { c_sel_end = index; break; } } /* Add one to start to skip over separator */ if (c_sel_start + 1 < c_sel_end) { textarea_select(ta, c_sel_start + 1, c_sel_end); return true; } return false; } J.

On Sat, 2012-06-23 at 02:11 +0200, Ole wrote: Committed as r14016. Thanks, J.