iconv_open behaviour on EILSEQ

[Stefan Hoffmeister]

: On Sat, 04 May 2002 21:21:10 +0200, Andreas Schwab wrote:

[quotes from POSIX 1003.1 snipped]

>This is pretty unambiguous, IMHO.  

I see your point.

I'd prefer some more explicit wording along the lines of "even if an
invalid byte sequence ... are the state variables updated." because in
an error case any callee-modified data typically is in an undefined
state.