Re: [Nmh-workers] Weird behavior with non-ascii code in headers

[David Levine]

> >Do we do any character replacement now, other than the
> >insertion of quotes around the display name that you mention
> >below?  If not, character substitution could introduce bad
> >behavior that we don't have now.
> 
> Actually ... yes!  If you use %(decode), we totally replace those
> characters and perform character set conversion.  We don't yet do that
> in the default component files when doing a reply, but it was actually
> my plan to make that happen for 1.6 (that requires some other stuff,
> like mhbuild doing RFC 2047 encoding).  Also ... if we cannot transform
> the characters to the local character set when doing character set
> conversion, the offending characters get replaced ... with a '?'.

I don't think that's a good idea.  Decoding and conversion should do
only what they're suppose to.  If they can't, they shouldn't produce
something different.  Esp. if the input is in error.  They should
flag the error and give up.

That malformed mail draft memo that you sent yesterday says this:

   o  Occurrences in addresses are especially problematic.  Agents
      supporting [EAI] may, if the 8bit conforms to 8bit syntax, elect
      to treat the messages as an EAI message and process it
      accordingly.  Otherwise, it is in most cases best to exclude the
      address from any sort of processing -- which may mean dropping it
      entirely -- since any attempt to fix interpret it definitively is
      unlikely to be successful.

David