Re: [Evolution-hackers] Bug #50046.

[David Woodhouse <dwmw2 infradead org>]

On Mon, 2004-01-05 at 09:10 -0500, Jeffrey Stedfast wrote:
> it's an empty group. it's syntactically meaningless. it's like extra
> whitespace between addresses... it's not needed.

It is not syntactically meaningless. It is precisely described by the
syntax.

I suspect what you mean is that it is _semantically_ meaningless, and
then you are only partially correct, since the semantics are dependent
on context.

In the context of the MUA looking at the headers and deciding which SMTP
RCPT TO: commands to issue, it is indeed semantically meaningless.

But these are user-supplied, user-visible headers, and RFC2822
specifically documents their use for conveying information other than
just the list of SMTP recipients' addresses.

The MUA has no business deciding that what I have typed for the benefit
of my recipients is semantically meaningless, and stripping it from my
mail. Likewise, if someone types such information into a mail which is
sent to me, my MUA has no business hiding that information from me.

The display-name of the group, even a group with no addresses, is just
like the display-name which indicates that fejj ximian com is actually
named Jeffrey Stedfast. Semantically meaningless when working out how to
send the mail, but definitely not semantically meaningless to the
_user_. Would you advocate that Evolution should strip those too, as
I've done above?

Think about usability. It's not sane to render everything in 'canonical'
form according to how an MUA would actually interpret RFC2822. There's
real person-to-person communication in there which you lose. Abide by
the principle of minimal munging, for your users' sake.

The same goes for the X-Spam-Report: headers which I posted an example
of. It's the same problem -- just because you show something which is
different only in whitespace or comments, that doesn't make it right.
You've lost or obscured information, needlessly. It's still wrong.

-- 
dwmw2