Danke für den Hinweis1 In MIME::Parser::Reader findet sich:
Quote# Parse up to (and including) the boundary, and dump output.
# Follows the RFC 2046 specification, that the CRLF immediately preceding
# the boundary is part of the boundary, NOT part of the input!
#
# NOTE: while parsing, we take care to remember the EXACT end-of-line
# sequence. This is because we *may* be handling 'binary' encoded data, and
# in that case we can't just massage \r\n into \n! Don't worry... if the
# data is styled as '7bit' or '8bit', the "decoder" will massage the CRLF
# for us. For now, we're just trying to chop up the data stream.
# NBK - Oct 12, 1999
# The CRLF at the end of the current line is considered part
# of the boundary. I buffer the current line and output the
# last. I strip the last CRLF when I hit the boundary.
Einem beendenden Boundary gehen also immer CRLF voraus. Demnach mache ich im Prinzip mit dem
truncate das Richtige.
Davon unabhängig gibt es noch das Problem CRLF-Übersetzung. Dazu werde ich noch ein wenig testen.