Tomi Ollila <tomi.ollila@iki.fi> writes: > When mail message is read from emacs, the message structure > obtained may contain parts which have content included > (`text/plain` for example) and other parts where content is not > included (`text/html` for example). > > In case content is included, the string is already available in > emacs' internal format and therefore mm-... functions should not > attempt to do further decoding for the data in temp buffer > provided for it. > > Currently when reply buffer is created, > notmuch-mm-display-part-inline () is used to provided quoted reply > content. This change makes the mm-... functions called by it use > 'gnus-decoded as charset whenever the content is already available. > > File .../emacs-23.3/lisp/gnus/mm-uu.el mentions: > "`gnus-decoded' is a fake charset, which means no further decoding." > --- > > I propose this patch is taken into 0.13.1. LGTM. I use it in my private branch. The bug fixed by this patch is really annoying. > > Please note that I'm not entirely sure my description above > is right. So those who knows more of these issues please check > my "facts". Your description matches with what I found in id:"871umc1int.fsf@steelpick.2x.cz". Thanks, -Michal