On Thu, Oct 13, 2005 at 11:12:47AM +0100, Dave Cridland wrote:
If you're really intent on having some mechanism to test for "raw"
features of the header content, then my personal view is that there
needs to be a "header" test, which normalizes whitespace, strips
header and footer, and decodes RFC2047, and a "rawheader", which
provides the header field value with all spaces, encoded words, etc,
intact.
Thinking about it, I suspected a new test might be useful in a
past mail of mine, too. So that may indeed be the way out.
The majority of people will intend that "banana" matches "Subject:
banana ", "Subject: =?Q?iso-8859-1?banana?=", and "Subject:banana"
and all other variations - the golden rule being if it looks as it it
ought to have matched when they observe it in the MUA, then it ought
to have matched.
Hmm. I can't think of anything that weakens that point. If MUAs would
display trailing spaces, things might be different, but they don't.
So you have me convinced: The header test should strip trailing white
space and next time I get hit by a spammer where this bites, I shall
think about specifying a new test operating on raw headers.
Michael