uuencoded attachments produce woe

[Matthias Andree]

On Sat, 07 Dec 2002, Allyn Fratkin wrote:

> Matthias Andree wrote:
> >The X is there to protect trailing spaces that might otherwise be
> >stripped during the transport.
> 
> ah, so it sounds like there are two major variants of
> uuencode (typical line):
> 1. ^M[^ ]{60}$
> 2. ^M.{60}X$
> 
> the first is pretty easy to recognize, the second is more problematic.

As you figured, it's not standardized, and I'd rather not rely on
uuencode being either of these; some put two spaces after "begin"
(remember the I-Love-You Signature that confused Outlook which would
uudecode without asking? There's a FAQ at Boris's site, piology.org),
some don't, and it magically sorts itself out as long as either of the
algorithms for decoding the 6-bit-sets holds, you can even or the lines
with 0x80 and most uudecoders will still extract the right stuff.

BTW, there's even a uuencode format that adds a per-line checksum
instead of the trailing X...

> i did find a website http://membrane.com/synapse/library/uuenc.html
> that indicates the character set used for uuencode so if we wanted to
> try for #2 we could.  unfortunately, since the letters A-Z and spaces are
> included, we might get some false positives.

Yes. base64 is hard, but as there is the MIME structure to support it,
we will be able to do that reliably some day. uuencode is harder,
because the mail structure doesn't support it and will never be able to
reliably figure that.