On Sun Jul 21 15:23 -0500 2013, Tomi Ollila <tomi.ollila@iki.fi> wrote: > On Tue, Jul 02 2013, John Lenz <lenz@math.uic.edu> wrote: > > > For my client, the largest bottleneck for displaying large threads is > > exporting each html part individually since by default notmuch will not > > show the json parts. For large threads there can be quite a few parts and > > each must be exported and decoded one by one. Also, I then have to deal > > with all the crazy charsets which I can do through a library but is a > > pain. > > This looks like a useful option. I just wonder what effect does different > charsets do to the output (is text/html content output verbatim (with just > json/sexp escaping of '"' -characters). > > If you added test(s) showing what happens with different charsets > (like one message having 3 text/html parts, one us-ascii, one iso-8859-1 > and one utf-8) that would make things clearer and (also) protect us from > regressions. > Ok, I'll add some tests, but everything is converted to UTF-8 by gmime. If you look, I didn't add any extra code actually. Instead I just changed the if branch taken depending on the option and the content type. The existing code already converted everything to UTF-8.