On 12/8/03 at 12:40 PM -0500, Keith Moore wrote:
On 12/8/03 at 10:43 AM -0500, Keith Moore wrote:
Nope, user agents SHOULD NOT decode RFC 2047 except for display purposes.
Do you consider sorting a "display purpose"? How about searching
given some user input?
seems fine to me.
Well, in that case I think you're being a bit naive when you say
earlier, "Decoding is only supposed to happen prior to display or
printing, not prior to storage." Requiring a search to do 2047 (and
therefore likely 822) parsing on each and every searched message in a
large message store is much too processor intensive in some
environments. Now, maybe that would require a shadow database of
pre-parsed messages along with the un-parsed 2047 fields to use in
replies, but that starts to become a significant burden too. In
either case, allowing the user to edit the subject of a reply
probably means de-coding and re-encoding the 2047, which runs into
exactly the kind of problem that Arnt was getting at in his original
message.
It would be nice to think that "decode only for display purposes"
would be a complete answer. It's not.
pr
--
Pete Resnick <http://www.qualcomm.com/~presnick/>
QUALCOMM Incorporated - Direct phone: (858)651-4478, Fax: (858)651-1102