At 6:55 AM +0100 12/1/04, Arnt Gulbrandsen wrote:
Sounds like an arms race.
1. Proxy adds header fields.
2. MUA learns to show only a select few fields.
3. Proxy sticks information into subject using []
4. MUA learns to suppress [].
In step 5, the proxy must stop using [] and switch to something it's
more difficult to suppress.
Possibly. Implementation details would matter significantly, I
think. In my mind, I would show these things where there was "room"
and not where there wasn't. So, message listings, I'd hide them.
Actual message display, I'd let them through. I think this would
meet most of the desires of the taggers, and so wouldn't be worked
around.
You can never distinguish a machine added [] tag from a parenthesis
inserted by a human. In fact, POPFile is a good example, because the
contents of the [] can be literally any string the user wants to insert
as a category label. So your MUA would have to query POPFile for the
list of category strings used or ask the user etc.
Again, implementation details matter here. If POPFile adds its tag,
my MUA processes the email with filters based on the tag (say, to ...
file ... it), but then hides the tag from the user in display (and
probably reply), everybody wins.
If a MUA did consider [] as a comment, wouldn't that break "Re[2]:
foo" parsing ?
Details, details. :-)
At 8:16 AM -0500 12/1/04, Nathaniel Borenstein wrote:
On Dec 1, 2004, at 1:46 AM, SM wrote:
The tags that are inserted in the Subject header break DomainKeys.
But not if the canonicalization algorithm is modified to ignore them.
Yup. Cooperate with the inevitable.
I'm much more used to wearing an implementor hat, where I have to
deal with the world as I find it, rather than a protocol designer
hat, where I get to fantasize about changing someone else's behavior.
That undoubtedly colors my outlook here.
--
Where are we going, and why are we in this handbasket?