In <200204301341(_dot_)g3UDfEe19994(_at_)astro(_dot_)cs(_dot_)utk(_dot_)edu>
Keith Moore <moore(_at_)cs(_dot_)utk(_dot_)edu> writes:
In practice, things that handle mail (like mailing list software,
gateways, virus filters, spam filters, etc. - certainly not just
MTAs) parse most of the structured fields for one reason or another.
Several of them also try to rewrite such fields or to correct
apparent errors in such fields - for instance, by taking strings
of illegal characters and quoting them according to RFC 2047
(regardless of where those characters appear in the header).
I tend to think that agents that try to correct "apparent" errors do more
harm than good, though they certainly seem to be the norm in the email
world.
On the Web, browsers that try to interpret malformed HTML have done untold
harm, because they simply encourage sloppily constructed web sites. Hence
the plethora of sites which say "this site can only be viewd in Internet
Explorer", or Netscape, or whatever. And decent browsers like Opera get
shunned because they refuse to play those silly games.
Usefor is quite strict on this issue. If any transport receives malformed
articles, either it passes them on *untouched* (most likely because it did
not bother to check for that particular abnormality), or else it drops
them on the floor *silently*. This is to ensure that every copy of a given
article that is found on the network is identical (a few "variable"
headers excepted). It was also done that way because transorts that tried
to "correct" things usually did so incorrectly.
This is a lovely theorectical discussion of how things could work. But before I
bother responding with my own take on this please show me the magic wand you
intend to wave that will undo 25 years and countless millions of lines of code
in common use today that don't work the way you think things should.
Without being able to effect such a change this sort of discussion is moot.
Ned