[Top] [All Lists]

Re: UUEncoding File Attachment

2010-01-28 22:55:04

Hector Santos <hsantos(_at_)santronics(_dot_)com> writes:
Getting rid of uuencode for MIME is not change for the sake of change.
uuencode was a badly broken way of handling attachments in multiple ways.
I say this as someone who has written his own uuencode and uudecode
processors and used the protocol for various applications, mail and
non-mail, over the years.  I'd be happy to never see it again.

It's like many pre-MIME ways of doing things that MIME now has better ways
of doing, such as some of the Usenet article signature algorithms.  The
old ways mostly work until you look at them funny or hit some corner case,
and then they turn into a mess.  It's that lack of attention to detail and
edge cases that I mean when I say poor engineering.  Most of those systems
were designed in a much simpler time and were, at the time, following the
"simplest thing that could possibly work" approach.  MIME is a refactoring
to deal with all of the interoperability problems that came up in practice
after using them for many years.

At the time the MIME specifications were being written, I was one of the few that was advocating at least a temporary way to handle UUENCODE within the MIME framework. The reason was to allow some greater form of interoperability, at least for the short term.

The arguments at the time were precisely what Hector mentions above. Co-existence of UUENCODE and Base64 I believe was one of the issues, but I am going by memory on this.

Being able to look back on this I have to admit that the correct decision was made NOT to include any support at all within the MIME framework from UUENCODED attachment types. Many of the messaging vendors (which my company at the time was one of), got around this problem in a very simple way. When a message came in, it was trivial to determine if it was MIME or non-MIME. If and only if it was non-MIME would we scan for any apparent UUENCODE blocks, thereby getting around potential MIME issues. In other words, having UUENCODE completely outside the MIME framework seems to have worked out best. And from the messages in this thread, it seems that all existing remnants of its use continue to be in the non-MIME context, so all is good. :-)

Best Regards,

<Prev in Thread] Current Thread [Next in Thread>