} Sounds like with this approach the caller should have to set the } right options for UTF-8-support on it's own and one has no influence } on how captured values are returned.
I'm not sure I understand the latter point. Are you saying you want more conversions to automatically convert captured values back to std::wstring type? This is all getting more complicated than I'd like this scheme to be. I'm definitely happier if you convert the input to utf8 before calling pcre, and converting the output from utf8 afterwards -- it seems like there less opportunity for trouble if pcre has a very clean encoding model. craig -- ## List details at http://lists.exim.org/mailman/listinfo/pcre-dev
