On 2009-03-17 at 20:39:50 [+0100], François Revol <revol@xxxxxxx> wrote: > B_UNICODE_ENCODING is actually UCS-2 (or maybe UTF-16, not even > sure...). > > Vision has a special case to handle B_UNICODE_CONVERSION as UTF-8 (it > just skips calling convert_*_utf8(), however this lets through invalid > UTF-8 strings. > > IMO we should support a B_UTF8_CONVERSION, rename B_UNICODE_CONVERSION > to B_UCS2_CONVERSION or whichever, to avoid misunderstanding, Sounds reasonable. > and > allowing the use of convert_ to also validate or eventually correct > broken strings by converting them from ISO latin1 as fallback (seems > ZETA's one does it when it finds broken UTF-8 as input). Not sure about this. Reporting an error and letting the caller decide what other encoding to try sounds better to me than hardcoding anything. Alternatively (or additionally) "lenient"/"do what you can" versions of the conversion functions could be added. CU, Ingo