--- In [email protected], "Sheri" <sheri...@...> wrote:
> The way you have proposed to do it (strings are returned as ANSI except when > the content has high value characters), any script that enables unicode > return-strings would need to test the content of each returned string to see > if it looks like "u\x53nnn". If the script wants to handle unicode, why not > let it declare that's what it wants and let it handle even low value > characters as unicode? Otherwise the same function called in a loop sometimes > returns a unicode handle and sometimes not! I accept your point. But should the new option mean that one returns _all _ results (excepting handles to com objects) as unicode strings? Or only those results that are meant to be BSTRs, leaving the VT_I4 and other numeric types alone? I assume the the latter.
