Note -- this isn't just -15, but -1 as well, and possibly others.
I have a utf8 text file that contains <c2 98> and <c2 80>. When converted to iso-8859-15 via:
iconv -c -f utf8 -t iso_8859-15//IGNORE input > output
The resulting file contains characters x98 and x80. These are considered invalid by some programs that expect iso8859-15 encoding -- including iconv itself.
Running the file through iconv a second time results in an error, or in the characters being removed if the -c or //IGNORE options are used.
As such, shouldn't the characters be dropped in the original conversion?